Beschreibung:
<jats:p> This study investigated the relationship between type, token frequencies, and graphemic context effects in predicting elementary students' pronunciations of vowel digraphs commonly used in reading tests. Three models were constructed to predict pronunciation responses: a final consonant model, a type-token model, and an invariant principal response model. For each of six selected vowel digraphs, six synthetic words were constructed according to orthographical constraints of English. Students of high and low reading ability in grades two, four, and six pronounced each of the 24 synthetic words. Results indicated the final consonant and principal response models to be superior. </jats:p>