Reading Patterns: Statistical Learning Mechanisms in Skilled Reading

The research included in the present thesis investigates how statistical learning mechanisms relate to the early stages of visual word recognition in skilled reading. Particularly, we ask whether the lifelong experience with textual regularities can provide a source of information for readers to exploit, and whether it can account for some of the fundamental skilled reading phenomena reported in the literature. Chapter 2 assesses the role of letter chunk frequency in morpho-orthographic processing, by comparing suffixes to highly frequent word endings in a masked priming lexical decision study. Chapter 3 includes this comparison as part of a hierarchy of contrasts, each of which tackles a linguistic feature potentially relevant for skilled reading. Selective discrimination responses are here captured through the adoption of the Fast Periodic Visual Stimulation paradigm (FPVS), coupled with MEG recordings. The FPVS technique also features the study described in Chapter 4, which reveals how the hidden regularities exhibited by a visual stream can inform the emergence of neural discrimination responses in skilled readers. Chapter 5 includes three experiments investigating whether the letter co-occurrence regularities inherent in the language can aid letter processing within briefly presented strings of consonants. Cumulatively, the evidence collected suggests that the knowledge of the statistical regularities that characterize the written input supports early orthographic processes in skilled reading.