The question we address is whether for a material similar to the one of print-to-sound associations, with many different alternative pronunciations for a given print unit, the performance is influenced by the consistency or uncertainty of the unit's association. The distinction is that the consistency estimate only takes into account the probability of the regular pronunciation of the spelling unit (or in a word, the average value across the word) when the other takes into account all possible pronunciations of a grapheme and their respective probabilities of association (as captured by an entropy measure). In this study, an implicit learning task was used in which the probability of occurrence of the next event as well as the number of alternative events that can happen next are manipulated. With this task, variations of performance as a function of the distribution of probability are found. This backs a previous finding of an effect of the consistency of the grapheme-phoneme associations (Lange & Content, 1999) and suggests that, contrary to what is assumed in the very influential model of reading of Coltheart and colleagues (2001), the knowledge that the reader encode about regularities is in the form of multiple associations rather than all-or-none grapheme-phoneme correspondence rules.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. (2001). DRC: A Dual Route Cascaded model of visual word recognition and reading aloud. Psychological Review, 108(1), 204-256.
Lange, M., & Content, A. (1999). Is printtosound conversion based on rules? Paper presented at the XIth Conference of the European Society for Cognitive Psychology, Gent (Belgium).