La compréhension des expressions argumentatives : la sémantique du prédicat aide à déterminer si un argument soutient un point de vue

old_uid4433
titleLa compréhension des expressions argumentatives : la sémantique du prédicat aide à déterminer si un argument soutient un point de vue
start_date2008/03/27
schedule10h30-12h30
onlineno
location_infoDépartement de psychologie, Salle 17
summaryModels of argument processing must explain how one determines whether a reason supports a claim. For example, the reason “many sentenced to death were found innocent” supports a claim about the immorality of the death penalty but not a claim about its ineffectiveness. One candidate for guiding this claim-reason connection is the thematic roles of the claim predicate. Ferretti, McRae, and Hatherell (2001) found that when reading a predicate of a statement (e.g., arresting), the features of thematic roles (agent-cop, patient-criminal) are immediately activated. Features of thematic roles should also be available during claim processing and allow readers to judge whether a reason supports a claim. In study 1, U.S. undergraduates listed features for argument predicates organized in thematic roles (agent: "someone who bans things is...", patient: "Something that is banned is..."). There was significant overlap of features across components of the predicates. For instance, the agent features of "ban" overlapped with the agent "permit" (knowledgeable, authority, powerful) but not "discourage" (mean, negative, rude, wrong). In contrast, the patient features of "ban" are more similar with "discourage" (wrong, harmful, dangerous) that with "permit" (acceptable, not harmful, interesting). These features of banning accurately captured that set of reasons found in the corpus analysis of warranted arguments. Such an analysis should help us develop a model of argument processing and help guide the creation of tutorials aimed at improving argumentation skills.
responsiblesRouet