This paper is based on thetopic of identifying preferences that a hearer utilizes to interpret a pronoun.It evaluates a coherence driven analysis based on grammatical role parallelism,thematic roles, implicit causality and subject hood. The first question ittries to address is the conflict between subject and grammatical preferenceswith respect to coherence analysis. Most of the work for this question is basedon Smyth and Wolf and the tries to find out whether supports their results or not.
Our results show that both preferences can be moderately when coherence iscarefully controlled for and argue that grammatical role parallelism is abyproduct of an independent interaction between information structure andaccent placement in a class of coherence relations. The second question it tries to answer isto distinguish between two concepts of thematic role preference which predictsa distribution of source and Goal interpretations and event structurehypothesis which predicts greater percentage of source interpretations inimperfective condition than in perfective condition. The results of Experiment2 supported the event structure biases involved in pronoun interpretation andare limited to those coherence relations which implicate even structure intheir formation.
The third question is what coherence driven theory predicts about incrementalprocessing and time taken to predict the pronoun.The author tries to tacklethis question by providing a analysis that captures documented preferencesbased on surface cues and also a range of phenomena that problematic for them.The paper then studies the phenomenon of implicit causality and its interactionwith coherence. The third experiment tests whether the biases found forimplicit causality verbs are a replication of Garvey and if it is a part ofexplanation relationship with responses with full stop condition.
the resultssupported the hypothesis. Also, the results represent one instance of amorecomprehensive setoff biases that drive predictive discourse interpretation. Finally, the grammatical subjectpreference is revisited and reasons to support against Crawley are analyzed.The author suspects there are sources of pronoun specific subject biases inpronoun interpretation they do not necessarily entail the existence of specialpurpose heuristic interpretation strategies instead prove to be better capturedwithin a more parsimonious expectation driven account.