Pronominalization and Expectations for Re-Mention: Modeling Coreference in Contexts With Three Referents

The relationship between pronoun production and interpretation has been proposed to follow Bayesian principles, combining a comprehender’s expectation about which referent will be mentioned next their estimate of how likely it is that potential re-mentioned using pronoun. Model received support from studies in several languages (English, Mandarin Chinese, Catalan, German), but tested contexts have limited two event participants, whereas natural language discourse often involves with more than participants. In this study, we conducted three story continuation experiments assess the performs complex contexts. Our results show even comprehenders can behave rationally when interpreting pronouns, they appear require sufficient context build up coherent representation situation do so. addition testing basic claim (Weak Bayes), test central prediction Strong form hypothesis: components model (next-mention expectations choice referring expression) are influenced by dissociated sets factors. comparison, Experiments 2 3 confirm closest fit Model, supports Weak Bayes, none our find evidence predictability affects pronominalization rates, corroborates Bayes. Finally, whether rate sensitive factors related ambiguity argument/adjunct status referents; participants vary pronouns most strongly based on grammatical role antecedent (subject or not), smaller effect presence/absence gender-matched competitor no syntactic position competing referent.