Direct magnitude estimation and interval scaling of naturalness and bizarreness of the dysarthria associated with amyotrophic lateral sclerosis

Scaling techniques are commonly used to judge the bizarreness (BIZ) and naturalness (NAT) of dysarthric speech. An equal-appearing interval scale (IS) is most commonly used. The validity of an IS is questionable because the nature (prothetic or metathetic) of the BIZ and NAT continua are unknown. Prothetic continua are quantitative (e.g., intensity) and metathetic continua are qualitative (e.g., pitch). Studies have shown that prothetic continua are not amenable to linear partitioning. The aim of this study was to determine if BIZ or NAT were prothetic or metathetic continua using a procedure outlined by Stevens (1975). The results of this study showed that, when magnitude estimates were plotted against intervals scores for both BIZ and NAT the relationships were curvilinear, suggesting that the continua are prothetic. However, such a strong conclusion cannot be drawn because several factors influenced the results. Logarithmic response biases, sequential contraction biases, and transfer biases may have resulted in a ceiling effect, creating the curvilinear relationship between the two sets of scores. The contribution of intelligibility tended to vary as a function of the scale and also the term. Further research is required to determine techniques that validly measure the BIZ and NAT of dysarthric speech and what perceptual dimensions relate to these judgments.