Recognition of phoneme invariance by beginning readers

This paper reports an experimental study of a possible confound in measures of young children's phoneme awareness — that of global similarity between words. We developed two otherwise identical versions of a test of phoneme invariance (typical item, ‘which starts the same asbeak, bowl orshed?’), one version controlling for global similarity and the other not. We administered both tests to 27 kindergarten children, along with three criterion measures of early literacy skill. Three converging results supported the importance of controlling for global similarity when attempting to measure phoneme invariance: (1) the subjects attained higher scores on the uncontrolled version, and half of those passing this version (11/22), were ‘false positives’ in that they did not pass the controlled version, (2) a metric of global similarity was significantly related to the proportion of children passing an item on both versions, and (3) there was a stronger relation between the controlled version and the criterion literacy measures of spelling and decoding than was found for the uncontrolled version. The educational implications of the results are discussed.