UC Berkeley

Abstract

Orthographic Learning Through Self-Teaching: Effects of Decoding Accuracy, Decoding Speed, Word Length, Morphology, and Individual Differences

by

Maria Deborah Callahan

Doctor of Philosophy in Education

University of California, Berkeley

Professor Anne E. Cunningham, Chair

The self-teaching hypothesis (Share, 1995) posits that readers learn the orthography of new words incidentally through the process of phonological recoding. In the current study, the

self-teaching hypothesis was tested by simulating everyday reading through the use of real and pseudowords, analyzing the effects of syllable length, and considering the independent contributions of general cognitive ability, including rapid naming ability and prior orthographic knowledge. A total of 94 second grade children were presented with 24 pseudoword targets over a 7-week period, each embedded 4 times within a short passage to be read aloud and independently. The targets varied by length (containing 1, 2, or 3 syllables) and by morphological legality (with real affixes or fabricated affixes). Findings indicated that the length of a word and its morphological legality significantly influenced decoding speed and accuracy. Shorter words were read faster and more accurately than longer words, although the difference in accuracy between 2- and 3-syllable words was only seen in the morphologically illegal condition. In all cases, morphologically legal words were read faster and more accurately than their morphologically illegal counterparts. A week after encountering a new pseudoword, the children were assessed for orthographic learning using two measures--an orthographic choice task and a naming task. Children demonstrated robust orthographic learning on the orthographic choice posttest for all words, regardless of word length or morphological legality, and for 2- and 3-syllable words on the naming posttest. There were no word length effects on posttest performance, however morphologically legal words resulted in more orthographic learning on the orthographic choice task than their illegal counterparts. In order to examine mediating factors that may contribute to children's orthographic learning, measures of vocabulary, prior orthographic knowledge rapid automatized naming, morphological awareness, and print exposure were administered. Children were also given tests to measure individual levels of vocabulary, prior orthographic knowledge, rapid automatized naming, morphological awareness, and print exposure. These factors, along with decoding speed, decoding accuracy, morphological legality, and word length were then analyzed for the contributions to orthographic learning using hierarchical linear and logistic regression models. Consistent with prior research, decoding accuracy and prior orthographic knowledge were the most powerful independent predictors of orthographic learning. The trajectories of decoding speed and accuracy across 4 exposures to a new word were also examined. It was observed that decoding duration time drops off dramatically after the first encounter and, intriguingly, that children tend to stay with their first pronunciation of a word, accurate or inaccurate.