How does case-mixing interact with lexical factors in visual word recognition? Implications for lexical organisation

This thesis investigates how different lexical and visual factors interact in reading, providing information on how the reading system is organised. Case-mixing and lexicality were found to be interactive in lexical and nonlexical naming in adults' reading but additive in children's. This supports the view that top-down lexical information helps experienced readers in overcoming the visual disruption of case-mixing. Children do not yet have sufficient stored lexical knowledge to enable such a mechanism to function effectively. Upper case disruption was smaller and equal across words and non-words. This lack of interaction occurs because unlike case-mixing, upper case does not disrupt attentional processes used in reading. Revisions of the top-down lexical feedback hypothesis are suggested such that lexical information feeds back to an attentional level of visual analysis. Inconsistencies in existing literature prompted investigation of the interaction between case-mixing and frequency. Early data was re-analysed and new experiments designed to favour the effect were run. Case-mixing was additive, not interactive, with frequency. This suggests that case-mixing disrupts an early stage in letter encoding before the construction of the code used to access the lexicon. The first evidence of an interaction was found. Smaller case-mixing effects on words from large neighbourhoods could also be explained by a greater degree of lexical activation feeding down and aiding words to overcome the disruption caused by case-mixing. Finally, an explanation for the disagreement within the literature on how frequency and regularity interact in children's reading was sought. A confound with age-of-acquisition (AoA) could be the key. When AoA was controlled regularity affected low frequency words only, but without it, regularity affected high and low frequency words equally. Implications that all four interactions have for the dual-route model of visual word recognition are considered.