Fernandes, E. 1, 2 , Luegi, P. 3, 4 , Correa Soares, E. 1, 5 , de la Fuente, I. 6, 1 & Hemforth, B. 6, 1
1 Université Paris Diderot
2 Labex EFL
3 Universidade de Lisboa
4 Centro de Linguística da Universidade de Lisboa
5 CAPES Foundation
6 CNRS
Co-reference is affected by the availability of pronominal forms in the language, but the interaction of language-based (global) exposure and (local) exposure, e.g., during experiments, is still unclear. We addressed this issue in questionnaires with Overt and Null pronouns in Portuguese sentences like The photographer greeted the journalist when he/Ø finished?, where he/Ø referred to either ?photographer? (Subject) or ?journalist? (Object).
While EP keeps a symmetrical distribution of Null and Overt forms, BP seems to be losing the Null forms. This would explain why EP shows the traditional ?division of labour? (Null retrieving Subject/ Overt retrieving Object) whereas BP shows no preference for Overt forms. If the processor is sensitive to pronouns? availability, skewing local exposure towards Null forms should elicit in BP the EP pattern. We created two Exposure conditions (50%/50%, 75%/25%) varying the relative amount of Null and Overt Pronouns. A significant interaction Pronoun:Exposure for BP confirmed our prediction: while Subject and Object were equally selected as the antecedent of the Overt pronoun in the 50/50 condition, choices for Subject and Object in the 75/25 condition were 30% and 70%, approaching the EP pattern. Our results indicate adaptation to the local statistical environment.