Unbounded dependencies have been one of the central concerns of both modern syntactic theory and psycholinguistics. The hope is that an understanding of how they are structured and processed will reveal essential features of the human language faculty as embodied in the grammar and likewise tell us much about the architecture of the human language parser. The purpose of this proposal is to increase our understanding of both the syntactic structure and the real-time processing of unbounded dependencies by studying them electrophysiologically, using event-related brain potentials or """"""""ERPs"""""""". Over the past decade, linguist have debated whether certain types of constituents (subjects, adjuncts, and even subcategorized constituents) form such dependencies at all and therefore """"""""leave traces."""""""" The role of specificity or """"""""D(is course)-linking"""""""" in determining the well-formedness of unbounded dependencies has also been investigated. Previous ERP research suggests that there may be a reliable electrophysiological index of the necessity to hold a filler in working memory pending its assignment to a gap: left-lateralized negativity appearing between 300 and 500 msec. Postimulus over anterior regions of the head, with a similar morphology and distribution to frontal negativity elicited by working memory tasks in other language contexts. The proposal is designed to take advantage of this finding by applying it to some of the theoretical issues outlined above. In particular, if left anterior negativity or """"""""LAN"""""""" reflects the role of working memory in language contexts, then it should be possible to ascertain (a) whether or not a filler-gap dependency is present in a given sentence and (b) to what degree this dependency is affected by other processing variables. Experiment 1 studies the formation of dependencies involving subjects. Experiments 2 and 3 test the hypothesis that adjuncts do not form unbounded dependencies or leave traces; Experiment 3 also examines whether fillers are associated directly with the verbs that subcategorize for them. Experiment 4 tests a prediction that unexpectedly filled gaps will result in a late positivity. Experiment 5 compares psychologically real traces with unfilled syntactic positions. Experiment 6 test predictions related to dependencies into specified NPs; Experiment 7 investigates the effects of specificity or D-linking in fillers on the processing of unbounded dependencies. There are two larger goals of this proposal. The first is to demonstrate that electrophysiological measures may serve as sources of evidence that can help to constrain linguistic theory. The second is in part methodological: validating the relation between an ERP effect and a specific mental operation. The proposal thus attempts to strike a balance between this methodological need and the application of what is already known about LAN to contexts where it may prove theoretically illuminating. As more is discovered about the nature of this component, it can be applied more incisively and reliably to further questions of purely theoretical interests, as well as to practical questions of how such structures, a universal feature of the world's languages, are acquired in child development and lost in the language breakdown of aphasia.
| Ueno, Mieko; Kluender, Robert (2009) On the processing of Japanese wh-questions: an ERP study. Brain Res 1290:63-90 |
| Cowles, H W; Kluender, Robert; Kutas, Marta et al. (2007) Violations of information structure: an electrophysiological study of answers to wh-questions. Brain Lang 102:228-42 |
