Acoustical Society of America
ASA/EAA/DAGA '99 Meeting
Lay Language Papers

Motor Theory of Melodic Expectancy

Frank A. Russo - russof@psyc.queensu.ca
Lola L. Cuddy
Department of Psychology, Queen's University
Kingston, ON, K7L 3N6, Canada

Popular version of paper 4aMU12
Presented Thursday Morning, March 18, 1999
ASA/EAA/DAGA '99 Meeting, Berlin, Germany

Despite the vast differences that exist in music across style and culture, certain similarities exist. Melody is one area where cross-cultural similarity can be found. There are patterns of melodic motion (i.e., how a melody proceeds from note to note) that appear to transcend many musical styles. Music theorist Eugene Narmour (1990) has forwarded a set of principles that describe these common melodic patterns. We submit that the origin of the principles and by extension the common melodic patterns may be vocal constraint.

Several experimental studies have demonstrated that listeners' melodic expectancies correspond with Narmour's principles (Cuddy & Lunney, 1995; Krumhansl, 1995; Thompson, Cuddy, & Plaus, 1997). More specifically, two sequential notes can set up common expectancies in listeners for what note will occur next. The melodic patterns outlined by each principle accurately describe these expectancies. The expectancies appear to be experienced in a similar manner by listeners of different cultures (Krumhansl, 1995).

Our experiments have shown that listeners judge melodies that adhere to principles as more cohesive than melodies that do not (Russo & Cuddy, 1996). This relationship has been demonstrated for British folk melodies, Chinese folk melodies, and 20th century atonal melodies (selected from Krumhansl, 1995).

We have found that adherence to the following three principles, as described by Narmour (1990) and quantified by Krumhansl (1995), is correlated with listeners' sense of cohesion in melody: 1) Proximity, tendency for small melodic intervals; 2) Registral Return, tendency for the third note of a three-note sequence to return to the pitch region of the first note; 3) Closure, tendency for phrase or melody to end with reversal in direction of pitch and/or a large interval followed by a smaller interval. Proximity and Registral Return contribute to our sense of cohesion throughout a melody. Closure however, only contributes to our sense of cohesion at the very end of a phrase or melody. The following images are examples of adherence and violation of each principle with corresponding audio (.wav) links.

The sound that is most familiar to humans and arguably the sound most important for melody is created by the human voice. In recent years, others and we have considered the voice as a possible source of the melodic expectancies considered here. It is our proposal that constraints of the human voice are manifest as expectancies that have influence over the composition, production, and perception of all melodies regardless of instrument or scale of implementation.

We define vocal constraint as the physical difficulty the average singer experiences in the vocal production of sound. Vocal constraint does not encompass difficulty that is perceptual or cognitive in origin. Proximity may be explained by the difficulty associated with singing large intervals accurately. Registral Return may be explained by the difficulty associated with accurately singing two sequential intervals that move in the same direction (due to the limits of the normal vocal range -- particularly if those intervals are large). Finally, Closure may be explained by a return to the pitch region that is most comfortable to produce vocally. We have found that endings of birdsong also tend to adhere to the principle of Closure (Russo, Sturdy, Cuddy & Weisman, 1996). Although songbirds and humans do not share a common evolutionary history, they may share a common constraint that is manifest in the song endings of both species as the principle of Closure.

There are at least three ways that vocal constraints may influence melodic expectancy. One way is through the natural evolutionary tendency for species to fit their environments. That is, our neural templates for organizing melodic information may have adapted to the constraints of the voice. A second possibility is that auditory templates reflect early learning in an auditory environment that emphasizes vocal communication (i.e., motherese). Finally, expectancies that resemble vocal constraints may be a vestige of the historical origins of music making (presumably vocal).

All three explanations for the influence of vocal constraints on melodic expectancies hinge on the assumption that vocal accuracy is consistent with melodic expectancy. We tested this assumption with twelve participants having diverse levels of vocal training (4 with no music training, 4 with non-vocal training, and 4 with extensive vocal music training). All participants were asked to sing three-note melodic sequences that were transposed into the center of each singer's comfortable range.

Results across training levels converged to suggest that singing accuracy was related both to 1) degree of adherence to the principles outlined above, and 2) perceptual expectancy as quantified by listeners' judgments in a previous empirical study (Cuddy & Lunney, 1995).

In a second experiment, we asked semi-professional singers to produce the same three-note melodic sequences that were used in the first experiment. Singing took place under two conditions. The first condition (mid-range) was identical to the first experiment in that patterns were produced in the middle of the comfortable vocal range. However, the second condition (high-range) transposed all sequences to the high end of each singer's comfortable range.

Singing accuracy in the mid-range was consistent with singing accuracy as assessed in the first experiment. However, singing accuracy in the high-range was not significantly correlated with perceptual expectancy.

A further point in support of the motor theory comes from analyses of principle adherence in real melodies. In particular, we have found no difference in adherence to principles in vocal vs. instrumental melodies. The lack of difference is suggestive since the physical constraints of most musical instruments differ from those of the voice (e.g., range).

In summary, the motor theory suggests that melodic expectancies that transcend musical style have origins in the constraints of the human voice. We propose that expectancies derived from vocal constraint are manifest in all facets of melodic processing (i.e., composition, production, and perception). Convergent evidence in support of the motor theory includes the correlation between vocal accuracy and melodic expectancy, and the similarity in adherence to principles within vocal and instrumental melodies. This evidence urges the next step: to assess the agreement between melodic expectancies and known physical constraints of the voice.


References
Cuddy, L. L. & Lunney, C. A. (1995). Expectancies generated by melodic intervals: Perceptual judgments of melodic continuity. Perception and Psychophysics, 57, 451-462.

Krumhansl, C.L. (1995). Music psychology and music theory: Problems and prospects. Music Theory Spectrum, 17, 53-80.

Narmour, E. (1990). The analysis and cognition of basic melodic structures: The implication-realization model. Chicago: University of Chicago Press.

Russo, F.A., & Cuddy, L.L. (1996). Predictive Value of Narmour's Principles for Cohesiveness, Pleasingness, and Memory of Webern Melodies. Proceedings of the Fourth International Conference on Music Perception and Cognition, 439-443.

Russo, F.A., Sturdy, C.B., Cuddy, L.L, & Weisman, R.G. (1996, August). Gestalt factors in the structure of birdsong and human melody. Canadian Society for Brain, Behaviour, and Cognitive Science, Montreal, Quebec.

Thompson, W.F., Cuddy, L.L., & Plaus, C. (1997). Expectancies generated by melodic intervals: Evaluation of principles of melodic implication in a melody completion task. Perception and Psychophysics, 59, 1069-1076.