Pronouncing printed words : investigating a semantic contribution to adult word reading

Abstract

When considering print-to-sound word reading, orthography and phonology are obviously involved. However, another system, that of semantic memory, might also be involved in orthography-to-phonology computation. Whether this occurs is debated in the literature both in the interpretation of behavioural results (e.g., Monaghan & Ellis, 2002; Strain et al., 1995) and in the implementation of semantic memory within computational models of word reading (Coltheart et al., 2001; Plaut et al., 1996). The central aim of this thesis was to investigate whether there is a semantic contribution to orthography-to-phonology computation in healthy adult word reading. Experiments 1-4 used a semantic priming design in which a picture prime was followed either two trials later (Experiments 1, 3, and 4) or one trial later (Experiment 2) by a word target, and this investigated priming of various word types. Regression investigations explored whether semantic features and imageability were unique significant predictors of ELexicon single word reading reaction times while statistically controlling for age-of-acquisition. The two ERP experiments (Experiments 5 and 6) investigated the neurocorrelates of imageability and semantic features and whether there are semantic effects early in the time-course of low frequency word reading. Experiments 1, 2, 4, 5, and 6 and the regression investigations show evidence of a semantic contribution to low frequency regular and low frequency exception word reading. There is also some suggestion of a semantic contribution to high frequency word reading (Experiment 2 and Regression analyses). From the results of the three lines of investigation, it is concluded that semantic information is involved in healthy adult word reading, and these results are best accommodated by the connectionist triangle model of word reading. These investigations also provide information concerning various word types and factors that contribute to “easy” and “difficult” words, semantic memory models and their accounts of priming, and the measures, age-of-acquisition, imageability, and semantic features.