[Comparison of different phonetic materials for perceptive analysis of dysphonia]

The authors describe a prospective study whose aim was to analyse the influence of the choice of phonetic material on the judgment of the panel. De Krom has already reported an experiment in which perceptive analysis showed that a sustained vowel was a useful material provided that the attack transitory was included (so called complete sustained vowel). He also showed that the panel gave more consistent results with speech or the complete sustained vowel than with the stable part of the sustained vowel. Our first objective was to find out whether we agreed with Krom's results, because for the study in the Dutch language to be applied to the French language, it was necessary to carry out an experimental validation given the cultural differences which might prejudice the results. In additional, we thought it likely that some acoustic information which was present in the attack transitories and was lacking in the stable part of the sustained vowel would run the risk of under-valuing the dysphonia in a study carried out on the stable part of sustained vowels. Our second objective was therefore to demonstrate that in cases of dysphonia, assessment based on the stable part of sustained vowels would be more lenient than those based on speech. The study was carried out using 80 voice samples from 60 dysphonic patients and 20 control subjects. The panel was made up of 7 experienced listeners. Agreement in the jury was judged by the percentage of identical responses given on 3 different presentations of each speech sample. This study showed no differences in the agreement of panel members over the three types of material by changing the severity of classification by the panel was analysed as follows: 1--Pearson r correlation coefficient, comparing on the one hand speech (the reference variable), and on the other samples of the complete vowel and the isolated stable part of the valve (tested variables). 2--A linear regression analysis. Our results confirm that assessments carried out on a sustain/vowel tend to cause a more lenient assessment than those carried out on speech. The assessment carried out on sustained vowels when the attack transient is included were of similar valued to those carried out using speech.