Scaling foreign accent: Direct magnitude estimation versus interval scaling

Scaling of degree of perceived foreign accent is common in studies of second language acquisition. Although interval scales are commonly used, it is not known whether accentedness is amenable to linear partitioning (i.e. described by a metathetic continuum) or whether accentedness is resistant to linear partitioning because it has degrees of magnitude or quantity (i.e. described by a prothetic continuum). This study determined if accentedness was a prothetic or metathetic continuum. Two groups of 10 native English listeners rated, in a self-paced experiment, degrees of perceived accent of native Italian speakers of English using both a seven-point equal-appearing interval scale and direct magnitude estimation (DME). DME scores were plotted against interval scores. This study found that accentedness (at least for Italian speakers) is amenable to linear partitioning. However, a wide dispersion of DME scores for '7' on the interval scale suggests a ceiling effect. A nine- or 11-point scale may better measure degree of perceived foreign accent. Individual listeners could reliably judge accentedness; however, inter-judge reliability was poor, possibly due to differences in listeners' internal standards of foreign accent or scaling artifact. Response biases occurred because foreign accent lacks a defined physical referent.