Mechanisms of spelling development

Learning to spell is a particularly challenging cognitive activity. This is in part because spelling in English is based on several levels of regularities that connect spoken language to print on the page. Consider the classic example of spelling of the word magician; its letters roughly reflect the word's component sounds, but they also reflect its units of meaning (e.g., c is used to represent the 'sh' sound in order to retain the root magic). Currently, we do not understand how it is that children develop sophisticated knowledge of the representation of units of meaning, or morphemes, in spelling, and yet it is clear that they are sensitive to the morphemic structure of word spellings at an early age. The set of studies outlined here will test a new theoretical account of how this might happen, in which I apply the ideas of statistical learning to spelling development Statistical learning is a concept that, over the last decade, has revolutionised the study of child development, and the time is ripe for its application to spelling development. In brief, I argue that children learn about the basis of spelling by attending to the frequency with which letters, sounds and meaning co-occur. For example, children might learn to represent plurality with the letter -s (as in the word final -s in dogs) through their experience of the simultaneous occurrence of the letter -s, the concept of plurality and the sound /z/ or /s/, rather than by learning a rule specifying that "all plurals are spelled with -s". The experiments described here will examine whether statistical learning of these co-occurrences can account for the timing and mechanisms of spelling development. The experiments provide an important test of the nature of statistical learning and, at the same time, they will assess the plausibility of a much-needed new model of spelling development.