Measures of the quality of spelling errors as predictors of later spelling performance

Spelling is a measure of the precision of lexical representations, which is important for word reading, comprehension, and writing. Performance on spelling tests is typically assessed in terms of correctness. We asked whether nonbinary measures of spelling errors provide additional information about lexical quality. Letter distance was the number of letters that needed to be changed to transform each error into the correct spelling. Phoneme distance was the number of changes necessary to transform the error into a phonologically plausible spelling of the target. Using longitudinal data, we asked whether these nonbinary measures of Time 1 error quality helped predict Time 2 spelling after considering the traditional correctness-based measure of Time 1 performance. We addressed this question by analyzing data from 756 children who received a standardized spelling production test at a mean age of 10.5 years and standardized spelling production and recognition tasks at a mean age of 15.9 years. The correctness-based measure of Time 1 spelling was a strong predictor of Time 2 performance, as expected, but letter distance had significant additional predictive value. Time 1 phoneme distance was not a significant predictor when correctness and letter distance were included in the model. The results show that we can gain information about lexicality quality from a spelling production test by using a nonbinary measure of spelling errors in addition to the traditional correctness-based measure.