We were interested in whether phonological awareness in young children emerges primarily as a result of lexical restructuring processes argued to be an integral part of language acquisition (Metsala & Walley, 1998). 'Lexical restructuring theory' proposes that segmental representations emerge primarily as the result of spoken vocabulary growth and associated changes in the familiarity of individual lexical items and inter-item phonological similarity relations. However, lexical restructuring theory as currently stated does not consider the nature of the phonological neighbours in the child's lexicon. For example, rime neighbours like pot, onset-vowel or lead neighbours like lock, and consonant neighbours like lit, are all considered equal neighbours of a target word like lot. However, given the abundant data demonstrating the psychological salience of the rime to young children (Goswami & Bryant, 1990, for review), it may be that many phonological neighbours in English are rime neighbours. If a prevalence of rime neighbours characterises the English phonological lexicon, then this might contribute to the developmental salience and utility of onset-rime representations in English.

We could find no analyses of phonological neighbourhoods in English in terms of type of phonological neighbour. We therefore analysed the corpus of single-syllable words in the Luce & Pisoni (1 998) database in terms of rime neighbours, onset-vowel or lead neighbours, and consonant neighbours in dense versus sparse neighbourhoods respectively, by type. This analysis showed that rime neighbours predominate in dense neighbourhoods. We then examined the effects of two factors on the development of rime awareness in young children, phonological neighbourhood density and sonority profile. Our experiment followed a 2 x 2 design, crossing sonority profile (good, poor) with neighbourhood density (dense, sparse). Five- and 6year-old children were given two tasks to measure awareness of the rime, the 'oddity' task and the same-different judgement task. The former measured awareness of the rime (e.g., pit, hit, got / meat, weak, seat / peak, dot, not), and the latter awareness of both rime (e.g., pit, hit) and coda (pit, got). In each task, simple CVC monosyllabic stimuli from high and low neighbourhoods were contrasted, as were stimuli with 1 good' or 'poor' sonority profiles. We found significant effects of both neighbourhood density and sonority profile on rime-level processing. Children were more accurate at making rime judgements about words in dense neighbourhoods and about words with 'poor' sonority profiles. However, there was no interaction between these factors. The results are interpreted with respect to developmental 'lexical restructuring theory' (Metsala & Walley, 1998).