Raghav Nathan '19
|
Title: Role of Referential Integration and Referential Ambiguity in Repeated Name Processing During Natural Reading
Problem: All previous work on the repeated name penalty during online reading has not used natural reading. Self-paced reading, RSVP presentation that is common in ERP studies, all rely on reading that isn’t natural. We wanted an ecologically valid assessment of the reading behavior that produces a repeated name penalty. The goal of this research was to explore the neural mechanisms of coreference using event-related potential and to understand the relative roles of referential integration and referential ambiguity in RNP. Activities & Interests: Raghav likes helping his FROGs, sitting on the Science Research couch every period of the day, and being the President of National Honor Society, President of Drama Club, Secretary of Musical Company, Vice President of Chemistry Club, Treasurer of Model United Nations, Treasurer of Debate Club, Chamber Choir and Choir Bass Section Leader, Varsity Tennis Captain, Varsity Swim.
|
|
Abstract:
The field of coreference processing is concerned with understanding how the brain processes words or phrases in a sentence when they refer to the same entity, that is, with the processing of anaphors (the word that refers to an entity previously introduced in the text) and antecedents (which is the word introducing the entity) by the brain. Two linguistic factors that impact coreference processing are the anaphor type - whether they are pronouns (such as “he” referring to “Clint” in a sentence) or repeated names (such as the noun “Clint” referring to a previously introduced “Clint” in a sentence), and the prominence of the antecedent - which refers to an entity's salience in the mental model of the sentence (when only “Clint” appears in a sentence this entity is more prominent compared to when “Clint” and “Nicole” appear in a sentence). The time taken to read sentences has been found to be adversely affected when sentences have repeated name anaphors with a prominent antecedent. This effect on coreference processing is called the Repeated Name Penalty (RNP). One hypothesis for the cause of RNP is problems with referential integration - integrating the repeated name anaphor with the entity requires more processing. Another hypothesis is problems caused by referential ambiguity - resolving the ambiguity introduced by repeated names impedes the integration speed.
In this research we investigate the role played by referential integration and referential ambiguity in RNP using ERP and eye tracking analysis. Referential integration is tested with N400 analysis, and referential ambiguity is tested with Nref analysis. We report what appears to be an early N400 effect that is smaller and more frontal than some of those previously reported. This is likely related to the combined eye tracking/ERP method that we're using. We also report that there was no Nref effect observed. We conclude that referential integration plays a dominant role in the processing cost associated with RNP.
The field of coreference processing is concerned with understanding how the brain processes words or phrases in a sentence when they refer to the same entity, that is, with the processing of anaphors (the word that refers to an entity previously introduced in the text) and antecedents (which is the word introducing the entity) by the brain. Two linguistic factors that impact coreference processing are the anaphor type - whether they are pronouns (such as “he” referring to “Clint” in a sentence) or repeated names (such as the noun “Clint” referring to a previously introduced “Clint” in a sentence), and the prominence of the antecedent - which refers to an entity's salience in the mental model of the sentence (when only “Clint” appears in a sentence this entity is more prominent compared to when “Clint” and “Nicole” appear in a sentence). The time taken to read sentences has been found to be adversely affected when sentences have repeated name anaphors with a prominent antecedent. This effect on coreference processing is called the Repeated Name Penalty (RNP). One hypothesis for the cause of RNP is problems with referential integration - integrating the repeated name anaphor with the entity requires more processing. Another hypothesis is problems caused by referential ambiguity - resolving the ambiguity introduced by repeated names impedes the integration speed.
In this research we investigate the role played by referential integration and referential ambiguity in RNP using ERP and eye tracking analysis. Referential integration is tested with N400 analysis, and referential ambiguity is tested with Nref analysis. We report what appears to be an early N400 effect that is smaller and more frontal than some of those previously reported. This is likely related to the combined eye tracking/ERP method that we're using. We also report that there was no Nref effect observed. We conclude that referential integration plays a dominant role in the processing cost associated with RNP.
| JSHS Winning Powerpoint |
