Using word frequency and parafoveal preview to determine the locus of contextual predictability and imageability effects: Evidence from eye movements during reading and lexical decision

The present thesis examines the time course of two semantic variables, contextual predictability and word imageability. Both variables can be said to reflect semantic aspects of meaning. For example the contextual predictability of a given target indicates the semantic context within which the target word occurs. The imageability of a given word reflects the meaning aspects of the word itself (Whaley, 1978). The word frequency effect (the faster response to commonly used high frequency words compared to low frequency words which occur less often) was taken to index the moment of lexical access (Balota, 1990; Pollatsek & Rayner, 1990; Sereno & Rayner, 2003) and by applying the logic of additive factors method (Sternberg, 1969a, 1969b), we determined whether the combined effect of each respective semantic variable was additive or interactive. This allowed us to examine whether there are semantic influences on lexical access. Previous research has been undecided and the question remains as to whether semantic variables operate during the lexical access processing stage, or alternatively after lexical access, for example in the post-lexical stage (e.g., Hand, Miellet, Sereno & O’Donnell, 2010; Sereno, O’Donnell & Rayner, 2006). Another aim of the thesis was to address the issue concerning the information presented to participants in the condition of ‘invalid parafoveal preview of a target’ (e.g., Sereno & Rayner, 2000). Several criteria were identified as being important in order to make the assumption that parafoveal processing was successfully inhibited on the pre-target fixation. Another aim of the thesis was to investigate whether word frequency and contextual predictability of the parafoveal word affected parafoveal preview benefit. Preview benefit was calculated by subtracting fixation durations in a condition of ‘valid’ preview of the target with an ‘invalid’ preview of the target. Experiment 1 utilised a lexical decision task to investigate the relationship betw