Fernandes, E. 1, 2 , Luegi, P. 2 & Teixeira, E. 3
1 University of Aberdeen
2 University of Lisbon
3 Federal University of Ceará
Comprehenders change the interpretation of Overt pronouns when these are less frequent in an experimental context. It has been argued that Null pronouns differ from Overt pronouns in having a stronger (Subject-)resolution bias. In Experiment 1, Brazilian and European Portuguese (BP, EP) participants (N=24, N=32) faced an environment with either the same amount of Null and Overt pronouns (equal distribution) or fewer Null pronouns (unequal distribution). Participants read sentences such as The athlete[Subj] consulted the orthopedist[Obj] at the hospital when he/Ø returned from the journey to Italy, and had to answer Who returned form the journey to Italy?. Speakers were less likely to choose Subject referents for Null pronouns in the unequal distribution condition.
In Experiment 2, EP participants (N=35) were first exposed to either Canonical or Noncanonical (in the language) disambiguation biases, and were then tested with ambiguous sentences. Participants were more likely to choose Subject referents for Null pronouns in the Noncanonical condition, against the exposure? statistics, but the probability of selecting each referent did not differ in the two conditions.
We show that Null pronoun resolution is also modulated by the statistical environment and suggest that resolution is guided by pronoun probability and expectations about referent probability.