Factors Affecting Comprehension of Ambiguous Three-Word Phrases: Local Semantic Constraints, Structural Preferences, Discourse, and Task Constraints

Daily language is full of ambiguity at various levels. While many have examined lexical ambiguity and sentential ambiguity, little research has examined ambiguity at the intermediate, phrasal level. We used Event Related Potential (ERP) to investigate ambiguity in 3-word phrases. Participants read four types of phrases: Ambiguous Unbiased (tall wheat farmer), Ambiguous Left (furry tiger painting), Ambiguous Right (slashed ship sail), and Semantically Unbiased (automatic printing machine). It was predicted that the unambiguous Semantically Unbiased condition would elicit less brain activity compared to other conditions because it is less semantically rich. At the 2nd word, the Ambiguous Right elicited a more negative N400 than the Semantically Unbiased whereas the Ambiguous Left did not show such an N400 effect. None showed a P600 effect. This suggests that the processor deemed the Ambiguous Right semantically anomalous at the 2nd word, but not syntactically anomalous. At the 3rd word, the Ambiguous Unbiased and Ambiguous Left stimuli elicited more negative N400 (250-550ms) and N700 than the Semantically Unbiased. Our interpretations are that the N400 reflects semantic ambiguity at the phrasal level and the N700 likely reflects the number of features needed to build and maintain a mental representation. Our findings support a semantic richness account of ambiguity processing at the multi-word level. This project also examines how these results change based on contextual factors such as task demands and sentential context