Resumen de The processing of anomalous sentences
The main goal of this thesis is to study the structural processing of anomalous sentences by the linguistic system. Investigating this question has both an empirical and a theoretical motivation. Empirically, we have aimed to shed light on how people comprehend sentences with novel, incongruent, or missing verbs (e.g., The waitress brunks the monk the ball) and sentences with anomalous word order (e.g., Gives the waitress the monk the ball), and whether they could be driven by input to produce structurally-dispreferred sentences (e.g., The dancer donates the soldier the apple). Theoretically, our experiments bear on the roles of lexically-independent and lexically-dependent syntactic information in language comprehension and production, on the number of stages of structural processing, and on the first stages of learning of verb subcategorisations. We approach the structural processing of anomalous sentences in 11 structural priming experiments. Empirically, our results suggest that comprehenders do construct well-formed syntactic representations for sentences with novel, incongruent, or missing verbs (and this is done via lexically-independent processing); that sentences with anomalous word order are comprehended by constructing syntactic representations which in part share syntactic information with the representations of similar but well-formed sentences; and that sentences with dispreferred verb-structure combinations are not produced due to the increased availability of abstract syntactic structure from exposure to well-formed sentences but only after exposure to similar dispreferred sentences with the same verbs as the ones selected for production. Theoretically, our findings imply that language comprehension is driven by lexically-independent syntactic processing, while language production, when the two types of processing are set in conflict, is driven by lexically-based processing; that the ordering of syntactic constituents is determined at a separate stage during syntactic processing; and that the learning of verb subcategorisations in adulthood is initially item-based, whereby constructions are linked to individual verbs.
