Modeling Lexical Access in Comprehension

Abstract

This thesis investigates lexical access by applying information-theoretic measures to quantify the complexity of verb processing in comprehension. It examines the effects of two key factors: Information (I), representing the complexity of an individual word form, and Entropy (H), representing the probabilistic variability within a word’s inflectional family. Using existing experimental data from neuro-typical participants and individuals with aphasia, the study employs both linear mixed-effects models (LMEMs) and non-linear sigmoid-based mixed-effects models to analyze the relationship between these factors and reaction times. The research addresses two main questions: to what extent do I and H affect lexical access, and how do these effects differ between neuro-typical and aphasic populations. The findings indicate that I and H have opposing effects on processing times, with I having an inhibitory effect and H having a facilitatory one. The models also reveal that the effect of H is significantly stronger than the effect of I and that this difference is even more pronounced in participants with aphasia. The non-linear models support these findings while also providing a more realistic framework that accounts for the bounded nature of human processing capacity. This work provides a deeper understanding of the "reduced ability" in aphasia, suggesting that it involves not just a general slowing but a specific alteration in how the cognitive system processes different types of information load during word retrieval.