Melodic and contextual similarity of folk song phrases

Various models of melodic similarity have been proposed and assessed in perceptual experiments. Contour and pitch content variables haven been favoured although music-theoretical and statistical variables have also been claimed to explain similarity ratings. A Re-analysis of earlier work by Rosner & Meyer (1986) suggests that simple contextual features can also be highly explanatory with more complex stimuli. A new experiment containing short melodic phrases investigated the effectiveness of several global and comparative variables. A multi-dimensional scaling solution indicated that both melodic direction and pitch range are highly relevant for making such similarity judgments and that the most salient aspects of melody when making similarity judgments are relatively simple context-dependent features.