Contrasting retrieval and encoding interference in reflexive pronoun resolution in EP and in BP

Paula Luegi, Márcio Leitão, Mário Carvalho, Matheus Barbosa and Armanda Costa- Universidade Federal da Paraíba

How to deal with contradicting results?

Goal & background: verify if syntactic principles restrict the antecedents of reflexive pronouns to thegrammatically possible ones, or if, instead, all referents are considered during reflexive resolution, contrasting: the Structure-based (Nicol and Swinney, 1989) and Unconstrained cue-based (Badecker and

Straub, 2002) accounts. We follow previous research that tested this hypothesis considering approaches based on relevant memory processes for language processing, namely encoding and retrieval (Laurinavichyute et al., 2017; a.o.): effects of ungrammatical referents might be explained by effects of interference during retrieval or encoding. While retrieval interference would reflect a violation of grammatical principles, encoding interference would be explained by effects of memory load. Distinction between encoding and retrieval interference might however only be established with the analysis of gender unmarked reflexives.

Experiments: two self-paced reading (moving-window) experiments, in EP and BP, with both gender

marked (“a si mesmo”) and gender unmarked (“se”) reflexives.

(1) O João/a Maria disse que o Pedro se magoou (a si mesmo)…

Results: Exp.1 (gender unmarked): effects, in both EP and BP, with higher reading times and lower

accuracy on question-answer on the gender-match condition (“João”…“Pedro”). Effects seem to be due to encoding interference. Exp.2 (gender marked): no effects of match or mismatch of referents in on-line measures (neither encoding nor retrieval interference), but similar effects (to Exp.1) in the question-answer task, with lower accuracy in the match condition. In this talk we will present some possible follow-up experiments to further research this question and to solve the contradictory results between Exp.1 and Exp.2.

---