Melodic expectancies and familiarization in music perception.

摘要

Leonard B. Meyer (1956, 1967) proposed that musical meaning arises from discrepancies between our expectations and the music we are listening to. It is claimed that these discrepancies still exist once the listener becomes familiar with a specific piece of music. It has been proposed that there are two different levels of expectancy, schematic and veridical, and that the persistent conflict between the two allows us to keep our interest even for music with which we are very familiar (Bharucha, 1987). There is no experimental evidence about the existence of discrepancies between schematic and veridical expectations. The present series of experiments attempted to fill this gap by investigating the effect of familiarization on different measures of melodic expectancy. The probe-tone technique (see Krumhansl, 1990) and a priming paradigm (see Bharucha, 1987) were used for three different types of melodies: British and Chinese folk melodies, and atonal melodies. The results obtained showed clear patterns of expectancy for specific notes in the cases of the British and Folk melodies, but not for the atonal ones. The next step was to have the listeners become very familiar with melodies that encompassed three levels of expectancy. A melodic recognition task was used to measure the listeners' memory for these melodies. Most participants were successful at memorizing them. After familiarization, there was experimental evidence from both types of expectancy measures that schematic expectancies were still present, regardless of the expectancy level of the note used during the familiarization phase. Namely, fit ratings remained higher and error rates remained lower for higher expectancy targets. In the results from the priming experiments, participants showed a bias to judge targets to be tuned when they were expected and mistuned when unexpected. There was also some experimental evidence for the presence of newly created veridical expectancies after familiarization. Ratings for the notes used during familiarization increased after familiarization. The reaction time and accuracy data showed only partial evidence of the presence of veridical expectancies, as significant changes in the data for the notes used during familiarization were only found for low expectancy targets. There were no instances where veridical expectancies became stronger than the schematic ones. Therefore, these results are in agreement with Meyer's theory. Results about the affective ratings showed that pleasantness ratings were closely related to fit ratings and increased after familiarization. Interest ratings were close to neutral and increased after familiarization only for low expectancy targets. The relationship between pleasantness and interest ratings varied with the expectancy level of the targets.