For skilled readers, word recognition is an apparently effortless cognitive process that can be swiftly performed across various presentation formats. A seminal study by Driver and Baylis (1995) investigated how two 'real-world' non-canonical orientations might disrupt visual word recognition of English words: namely, 90-degree rotation (e.g., the title on the spine of a book) and a vertical letter arrangement from top to bottom (e.g., the title on a building's marquee banner). Driver and Bayliss found that rotation had a less detrimental impact on speed and accuracy than the marquee presentation, which they interpreted as evidence for the importance of preserving a word's 'principal axis'. The current study seeks to replicate the findings of the original experiment and include a canonical (typical) presentation, which was absent in the original work. This 'baseline' inclusion allows for a clearer assessment of the effects of non-canonical orientations on typical lexical processing in visual word recognition. Additionally, we systematically examined the potential attenuation of these orientation manipulations on two key variables relating to word items (frequency) and nonword items (pseudohomophony) respectively. Our findings broadly replicate those of Driver and Baylis (1995), indicating a graded impact of non-canonical presentation (canonical > rotated > marquee). Lexical effects remained remarkably robust across all presentation formats, suggesting that access persists even when letter strings are presented in such unfamiliar orientations. We interpret this robustness as indicating that the effects of non-canonical presentation on lexical processing are quantitative rather than qualitative in nature.
Allen et al. (Tue,) studied this question.