Co-occurrence statistics affect letter processing

Reading requires the successful encoding of letter identity and position within a visual display, a process that relies on both visual and linguistic resources. In a series of experiments, we investigate whether readers’ lifelong experience with letter co-occurrence regularities supports letter processing. Skilled readers were briefly exposed to strings of five consonants; critically, letters in position 2 and 4 were embedded in either high (B in MBL) or low (PBG) transitional probability (TP) triplets. When presented with two strings differing by the critical letter (e.g., MBLSD vs. MCLSD), participants correctly identified the right option more often in high-TP than low-TP contexts, regardless of position. Experiment II featured both a Same-Different and a Reicher-Wheeler task with response time constraints, and further qualified the contextual facilitation effect, with high-TP eliciting faster ‘same’ judgements only for letters in position 2. In a third experiment, context had no effect on Same-Different matchings with strings of pseudo-characters sharing letter low-level visual features. Our results indicate that co-occurrence statistics affect letter recognition in tasks that emphasize whole-string processing. This effect is genuinely orthographic, as it is conditional on intact letter identities, and with increasing task demands it only surfaces for letters close to word onset.