An ERP Investigation of Semantic Richness Dynamics: Multidimensionality vs. Task Demands

Abstract

Semantic richness is a multidimensional and dynamic construct that can be defined as the amount of semantic information a word possesses. In this thesis, the semantic richness dimensions of number of associates, number of semantic neighbours, and body-object interaction were investigated. Forty-eight young adults were randomly assigned to perform either lexical decision (LDT) or semantic categorization tasks (SCT). The goal of this thesis was to investigate behavioural and electrophysiological differences (using the Event-Related Potential technique) between semantically rich words and semantically impoverished words. Results revealed that the amplitude of the N400 ERP component was smaller for words with high number of associates and high number of semantic neighbours compared to words with low number of associates or low number of semantic neighbours, respectively, but only during LDT. Behavioural results, however, only showed an accuracy and reaction time advantage (during item analyses) for words with many associates. In contrast, N400 amplitudes did not differ for words with high body-object interaction rating when compared to words with low body-object interaction rating in any of the tasks, although a behavioural reaction time advantage was observed in item analyses of the LDT. These results suggest that words with many associates or semantic neighbours may be processed more efficiently and be easier to integrate within the neural semantic network than words with few associates or semantic neighbours. In addition, the N400 effect was seen in the LDT but not in the SCT, suggesting that semantic richness information may be used in a top-down manner in order to fulfill the task requirements using available neural resources in a more efficient manner.