LingLunch : Hans Wilke

Jeudi 19 Mai 2022, 12:00 to 13:00
Karen De Clercq et Ira Nowek (LLF)

LLF – Bât. ODG – 5e étage – Salle du conseil (533)

Hans Wilke (University of Edinburgh)
Given-before-new, and then what? Information structure and sentence structure influence processing and expectations

Comprehenders have expectations about how information will be presented in a sentence and make predictions about what will be mentioned next (Cristea & Webber 1997; Roland et al. 2012). In a series of experiments, we test how information-structural features and sentence structure influence both. While there is extensive evidence that comprehenders prefer given information to precede new information (given-before-new principle: Gundel 1988), this principle has been primarily tested by considering the information-structural features encoded in syntax (e.g., definite NPs represent given information, indefinite NPs represent new information). The current project revisits the given-before-new principle with a novel approach: focusing on the discourse-status of information and disentangling new-/givenness from syntactic structure.  We carried out self-paced reading experiments with multi-clause sentences, of which only one clause contained given (discourse-old) information. A preference for given-before-new is found, as well as an interaction effect between information status (given/new) and clause ordering: given-before-new information ordering had a greater effect in sentences where the subordinate clause came before the matrix clause, than vice versa. Following this, I will present a selection of experiments that show how the information-structural principle of at-issueness influences expectations about what will be mentioned next. At-issueness concerns itself with what the main point of an utterance is (Potts, 2005; Koev, 2018) and consequently, what is felicitously available to be picked up in subsequent discourse (Hunter & Asher, 2016). For multi-clause sentences, clause type (matrix /subordinate) and clause position (sentence-early/sentence-final) have been posited to contribute to at-issue status (Jasinskaja 2016; Syrett & Koev 2015). We measured reading times when a pronoun with two possible antecedents is disambiguated to a referent in either a matrix clause or a subordinate clause, in sentence-early or sentence-final position. We find that clause position impacts reading times, but that the effect of clause type depends on the type of subordinate clause comprehenders encounter.