Fast and specific access to orthographic knowledge in a case of letter-by-letter surface alexia

A series of experiments assessed priming for single letters and words in a letter-by-letter reader (IH) when primes were displayed briefly (between 100-500msec) and masked. Consistent with previous claims that letter-by-letter readers have difficulties accessing orthographic letter codes, IH failed to show normal cross-case priming for single letters in a naming task (e.g. a/A). Nevertheless, IH showed robust cross-case priming for four-letter words that have few if any perceptual features in common between upper and lower case (e.g. read/READ; the letters r/R, e/E, a/A, and d/D are visually dissimilar in lower/upper case), even at prime durations that failed to support priming for single letters. Furthermore, priming extended to pseudowords (e.g. DEAT), and was highly specific given that no priming was obtained between orthographic neighbours (e.g. face did not prime FACT). Based on this pattern of results, we argue that IH gains relatively normal access to orthographic representations, and that his letter-by-letter reading reflects a partial disconnection between orthographic and phonological representations. Within the context of a disconnection account, we provide an explanation of the paradoxical finding of robust word priming in the absence of single letter priming.