Communicating voice emphasis in harpsichord performance

We investigated the means used by harpsichordists to communicate voice-specific melodic emphasis. Twelve harpsichordists were asked to perform a short Baroque polyphonic piece by Frescobaldi (1583-1643) on a harpsichord equipped with a MIDI console. Three conditions were tested, each requiring performers to emphasize a particular voice in the piece (soprano, alto, tenor). Four parameters were analyzed: velocity, note onset asynchrony, timing deviations, and articulation. Mean onset asynchronies were much larger than those observed in organ or piano performance, averaging more than 100 ms between outer voices. However, asynchronies did not vary significantly between conditions, suggesting that they do not play a major role in communicating voice emphasis. On average, notes belonging to the upper three voices were struck with a higher velocity and played with a more detached articulation when they were emphasized. Although timing deviation patterns were globally similar across all conditions, we observed local differences that corresponded to passages in which a specific melodic gesture was accentuated by means of ritenuto (tempo deceleration).