From novel to familiar: The learning of pitch intervals and event frequencies in microtonal music system

This PhD project aims to investigate the learning of novel music systems during brief exposures (up to 20 minutes). Specifically, it examined the learning of microtonal pitch systems, including the learning of their pitch interval structure and the frequency of occurrence of each pitch member of the systems. Five experiments have been conducted and the findings are separately submitted to peer-reviewed journals for publication. The submitted manuscripts are provided as individual chapters in this thesis. Experiments in MS 1 explored the influence of structural similarity on the learning of novel tuning systems. Non-musicians were exposed to microtonal tuning systems that are different from the western tonal system (12-tone equal temperament or 12TET) in terms of temperament (equal or unequal) and pitch intervals (based on frequency ratio(s) or frequency difference). It was hypothesized that systems that are structurally similar to 12TET in terms of temperament and pitch interval definitions (such as 11TET) would be learned more rapidly than those that are dissimilar (such as the 81 primes scale). Participants were tested on their memory of the relative pitches of each tuning system as well as their knowledge of which relative pitches correspond to which tuning system (pitch membership). Results indicate better pitch memory for the tuning system with two frequency ratios (Tuning system: WF) while learning of pitch membership was not found after a brief exposure.;MS 2 targeted the incidental learning of a microtonal scale that follows a pitch interval structure similar to WF in MS 1, which is unequal tempered with two frequency ratios. This MS extends the data-driven interpretation of MS1 in terms of my overall hypothesis. Thus, a new paradigm is developed and learning of the scale is implied by the level of sensitivity to the incongruent pitch intervals. This paradigm utilises a timbre shift detection task where participants are required to detect a timbre shift in one of the pitches of a microtonal melody. The pitch before the shift is manipulated to be either a member of the scale (congruent), just like the rest of the pitches in the melody, or a pitch from a different musical scale (incongruent). The melodies are arranged such that an incongruent pitch is a statistical cue to an upcoming timbre shift, so that when participants hear an incongruent pitch, they would have high expectation of the imminent shift and therefore respond faster when it comes. However, faster reaction times would only be observed if participants have sufficiently learnt the pitch intervals of the microtonal scale and the relationship between the incongruent pitch and the timbre shift. In other words, a faster reaction time is expected in the condition where an incongruent tone was placed before the timbre shift, than in the condition where the incongruent tone was absent. Non-musicians in MS 2 showed successful learning of both, and this experiment was extended in MS 3 where musicians were tested. Surprisingly, the same results were not observed in musicians who are professionally trained in 12TET only (general musicians) and those who are microtonal music experts (trained in 12TET and also had extensive experience in multiple microtonal tunings). While similar results with those from the non-musicians might be obtained by having a comparative sample of general musicians, the unexpected results among microtonal musicians, ironically, might still be due to their musical experience and practice.;MS 3 also interrogated our overall hypothesis further by examining the learning of the frequency of occurrence of pitches of the same microtonal scale (the relative event-frequencies) in non-musicians. The effect of structural similarity was once again investigated by the manipulation of event-frequency of the microtonal scale so as to be similar or dissimilar to the functional hierarchy (difference in usage) of pitches in diatonic music. Participants were exposed to melodies generated based on the similar or dissimilar pattern of event-frequency and goodness-of-fit ratings of pitches of the scale were obtained at random points in the melodies. Results delineate an insignificant effect of structural similarity, since learning was found in both patterns of event-frequency, implied by the positive correlation between the goodness-of-fit rating and the frequency of occurrence of the corresponding pitches, and demonstrated by an analytical model of the results showing the influence of the pitch-class probabilities. This result illustrates the ability to learn event frequencies in a novel musical scale among musically untrained participants. Altogether, the findings in all the experiments indicate that non-musicians can learn aspects of the pitch intervals and event-frequency of an unfamiliar microtonal scale rapidly, and learning is faster when the scale is well-formed and octave-based with two frequency ratios. Inconsistent with predictions, general and microtonal musicians did not show such rapid learning, and how this might relate to the paradigm used is discussed. This project provides informative understanding of how listeners with different musical background and experiences approach and learn a novel music system, which has implications for research in music cognition and perception, traditional and computer music composition, music production, and music performance including improvisation.