The proposed research aims to understand written and spoken word comprehension via computational, behavioural, and electrophysiological (EEG/ERP) investigations.
Computational Investigations. I will develop improved connectionist models which capture the time-course of interactions among orthography, phonology, and semantics during word comprehension. These models will address key limitations of past work and make contact with additional methodologies (e.g. EEG). This will involve extending my connectionist models of written word comprehension and decision making (Armstrong, Joordens, & Plaut, 2009; Armstrong & Plaut, 2008) to include representations of spoken words activated by auditory input over time, and representations of written words activated by a visual input that arrives all at once. These models will be used to simulate performance in a popular task in the literature—lexical decision—and form a unified and explicit computational framework to capture a range of previously-reported behavioural and EEG/ERP effects. The emergent characteristics that arise when orthography, phonology, and semantics interact are predicted to address several criticisms of past models (e.g. absence of stage-like effects).
Behavioural/EEG Investigations. A secondary component of this project consists of behavioural investigations of visual and auditory lexical decision that will produce data against which the models can be compared. The main behavioural experiments will assess performance in adults. Pilot studies that include EEG recordings and child participants will assess the feasibility of extending the main study to these domains.
I will develop and release a public implementation of my empirically-validated models and data. These models will account for a broad range of existing effects, generate predictions for new experiments, and be readily extendable to related areas (e.g. language impairments, language learning).
Call for proposal
See other projects for this call