"Nasalance" vs. listner judgements of nasality

In this article, judgements of nasality from sound field recorded utterances of 23 children with repaired palatal clefts are compared with "nasalance" values derived from TONAR II analyses of parallel sound separated recordings. For the nasality judgements, the sound field recordings were presented in forward reproduced mode to 20 "naive" listeners and in backward reproduced mode to 10 naive listeners. A four-level task progression was followed in each mode of presentation: sorting the responses by "normal" or "abnormal" nasality, ranking them by severity of nasality, classifying them within five degrees of nasality, and, finally, assigning a discrete score to the magnitude of nasality perceived in each recording. Scores from the individual judges were highly variable, especially in the first listening task. They were consitently more variable when nasality was judged from backward reproduced recordings. As the listeners gained experience and as the listening tasks demanded more precision in nasality judgements, variability was reduced. Correlations computed between the nasality ratings and nasalance measurements increased and variability decreased. Highest agreement between physical and perceptual measurements was found when variability among the judges was reduced by pooling the scores and comparing the mean scores of nasality with the nasalance scores. Using this procedure, a correlation of .91 was obtained between listener judgemental scores of forward reproduced speech and nasalance scores. Under these conditions, the instrumental score could apparently account for over 80 per cent of the variability in the listener judgements of nasality. From the findings of this study, it was concluded that listeners without prior training are capable of judging nasality reliably and that nasalance scores provide a valid correlate of perceived nasality. The use of backward reprodeced recordings was questioned as a means of assessing nasality.