Production schemas and motor programs in melodic singing.

Abstract

To produce a melodic sequence of notes, a singer must construct and execute an articulatory program that controls the vocal movements for the sequence. This present study investigates the nature of these programs and their musical features. We extend research on music perception to music production. Expert singers participated in four experiments with a response-priming (RP) procedure adapted from work by Meyer and Gordon (1985, 1987). On each trial of the procedure, a subject prepared to sing a primary melody composed of five notes that formed a short tune. After this preparation, the subject either sang the primary melody as rapidly as possible, or else switched and sang another secondary melody instead. An auditory response signal cued which melody to sing. Subjects' response latencies, response durations, and error rates were measured as a function of various musical relationships between the primary and secondary melodies. For the secondary melodies, lesser latencies, durations, and error rates occurred when the secondary melody had the same initial note as the primary melody. Latencies and error rates significantly increased when the singers had to change keys between the primary and secondary melodies. Exact transpositions between primary and secondary melodies also yielded marginally lower latencies and error rates. We conclude that articulatory motor programs for short melodies are represented as sequences of exact upward and downward steps from a particular initial note. Contrary to some theories, the qualitative contour of a melody is not included, per se in this representation. Also melodic keys appear to be arranged within the minds of expert singers according to relationships modeled by the musical circle of fifths. The closer on the circle of fifths two keys are, the more notes they share. These shared notes tend to ease the switch between keys, with "stronger" tonal-note functions yielding shorter latencies and lower error rates.