Congruency and typicality effects in lexical decision

Abstract

This thesis describes basic research into visual word recognition and decision making. Determining the best matching lexical representation for a given stimulus involves interactions between representations. The standard task for studying these processes is the lexical decision task (LDT), but there is still debate regarding the factors that affect how individuals make lexical decisions. The nature of lexical interactions and the processes underlying lexical decision-making were addressed here by testing response congruency effects in the masked priming variant of the LDT. The results of seven masked priming experiments showed a robust response congruency effect that depends on the difficulty of the word-nonword discrimination. This finding resolved apparent inconsistencies in previous research. The experiments were simulated using the Bayesian Reader and the Spatial Coding Model (SCM). The probability based Bayesian Reader model failed to accommodate the findings. However, a good fit to the data was provided by a modified version of the SCM in which the assumptions regarding the nature of lexical interactions were changed such that word nodes inhibit only (closely) related competitors. The model also assumes that the difficulty of the word-nonword discrimination affects the degree to which stimulus typicality informs lexical decisions. A critical issue for these experiments involved the definition of orthographic typicality. An algorithm for measuring orthographic typicality and for generating nonwords with a specific level of orthographic typicality (OT3) was developed. An unprimed LDT experiment showed that OT3 affected decision latency even when other standard measures of orthographic typicality were controlled. Two additional masked priming experiments showed that highly typical primes lead to faster word responses and slower nonword responses than less typical primes. Overall, the results of this research enhance our understanding of the processes underlying visual word recognition and lexical decision making, and also have important methodological implications for the field.