The role of semantics in early reading

Abstract

The aim of the dissertation was to obtain insight in semantic effects in word identification of Dutch children. Chapter 2 focused on the predictive value of risk factors, cognitive factors, and teachers' judgments in a sample of 462 kindergartners for their early reading skills in Grade 1. The discriminatory power of all predictors appeared to be insufficient. Results of a discriminant function analysis, however, demonstrated that a combination of different factors increased the accuracy of prediction. Chapter 3 describes the results of two intervention studies in Grade 1. In the first study, 121 poor beginning readers were assigned to a semantically-oriented training, a phonologically-oriented training or a control group. At post-test, children in both experimental training programs showed similar gains in word-identification skills. In the second study, the experimental-training programs were modified and extended. About 83 poor beginning readers participated in one of both experimental groups or were assigned to a control group. Result showed that all groups performed statistically equally across all reading measures. In Chapter 4, the relationship between semantic skills and word-decoding skills was examined. In Experiment 1, results revealed no differences between poor decoders and good decoders in word-association skills, whereas poor readers were more error prone in a semantic-categorization test. In Experiment 2, children from Grades 1 to 6 participated in two types of semantic-categorization tasks and a word-decoding test. Poor readers showed longer reaction times on both types of categorization tasks than average readers and good readers. Chapter 5 focused on imageability effects in isolated-word reading in Dutch children from Grades 2 to 6. Word-reading skills were assessed by lexical decision (Experiment 1) and naming (Experiment 2). Results of Experiment 1 revealed an imageability advantage in the non-speeded task in latencies. In addition, in low-frequency words, more errors were made in low-imageability (LI-) words than in high-imageability (HI-) words. Accuracy analyses revealed a significant grade by instruction by imageability interaction; in the higher grades, only in the speeded test more errors were made in LI-words than in HI-words. Results of Experiment 2 indicated that in naming, no imageability advantage was demonstrated for accuracy scores