Causal inferences and expectations in discourse processing

Thesis

Successful language comprehension requires comprehenders to understand not just individual linguistic elements, but also the relationships between them. One set of relations central to discourse comprehension is causal coherence relations, such as Result (cause-effect) and Explanation (effect-cause). This thesis investigates the establishment of implicit Explanation relations, a subtype of causal relations, in discourse processing.

Prior studies have demonstrated that local linguistic cues can guide comprehenders to infer and expect Explanation relations in discourse processing. For example, when reading the sentence Jenny fired the guy who had been embezzling money, comprehenders will infer an Explanation relation between the firing event and the embezzling event. Even before they reach the relative clause, they may expect an upcoming Explanation that explains the firing event. However, significant questions remained about what linguistic elements constrain implicit Explanation relations, how comprehenders infer such relations across different discourse contexts, and where in the linguistic structure these relations can be established.

The present thesis addresses these gaps through experimental methods and formal semantic analysis. A series of studies examined both inter-clausal and clause-internal Explanation relations using comprehension tasks, self-paced reading, and eye-tracking-while-reading. The findings show that discourse expectations for upcoming Explanation relations are constrained by syntactic embedding (e.g., islands) and discourse structure (e.g., right frontier). Furthermore, the present thesis shows that while Explanation relations can be established within a clause through cues such as resultative adjectives (e.g., The broken window got struck with a stone. ⇝ ‘the window was broken because of the stone.’), clause-internal relations are more fragile and delayed compared to inter-clausal relations. Together, these findings advance our understanding of discourse processing by shedding light on how and when readers establish implicit Explanation relations during discourse comprehension.

Authors: Yao, R

• DOI: 10.5287/ora-vjy1g0gmx
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: sentence processing; discourse coherence; preidtive processing; discourse structure; discourse processing; causal inference
• Subjects: Semantics; Linguistics; Psycholinguistics