(Dis)agreement on Sight-Singing Assessment of Undergraduate Musicians

Assessment criteria for sight-singing abilities are similar to those used to judge music performances across music school programs. However, little evidence of agreement among judges has been provided in the literature. Fifty out of 152 participants were randomly selected and blindly assessed by three judges, who evaluated students based on given criteria. Participants were recorded while sight-singing 19 intervals and 10 tonal melodies. Interjudge agreement on melodic sight-singing was tested considering four items in a five-point Likert scale format as follows: (1) Intonation and pitch accuracy; (2) Tonal sense and memory; (3) Rhythmic precision, regularity of pulse and subdivisions; (4) Fluency and music direction. Intervals were scored considering a 3-point Likert scale. Agreement was conducted using weighted kappa. For melodic sight-singing considering the ten tonal melodies, on average, the weighted kappa (κ_w) were: κ1_w = 0.296, κ2_w = 0.487, κ3_w = 0.224, and κ4_w = 0.244, ranging from fair to moderate.. For intervals, the lowest agreement was kappa = 0.406 and the highest was kappa = 0.792 (on average, kappa = 0.637). These findings light up the discussion on the validity and reliability of models that have been taken for granted in assessing music performance in auditions and contests, and illustrate the need to better discuss evaluation criteria.