This article attempts experimentally to confirm the Grouping
Preference Rules (GPR's) proposed by Lerdahl and Jackendoff in A
Generative Theory of Tonal Music, as well as to develop an ordering of
salience for these preference rules. Two experiments are performed. In
the first, subjects (both musicians and nonmusicians) heard passages
from preexisting music and were in some cases provided with a context
for the examples. The subjects were asked to place segmentations among a
line of dots which represented the notes. In this manner, the validity of
the GPR's could be tested. A second experiment put GPR's into
conflict with one another. A series of artificially generated descending
and ascending major and minor scales was presented. Two different GPR's
for each example would place segmentation at different places in the
example. Listeners were forced to choose between the two by placing only
one segmentation in a line of dots. The main conclusions of this study
can be summarized as follows:
- Lerdahl and Jackendoff's GPR's are valid, both for musicians and
- Musicians were more "efficient" in their placement of segmentations,
due to their familiarity to the examples heard or to their familiarity
with music in general.
- The GPR's utilizing proximity of attack (GPR 2), timbre (GPR 7),
and change of dynamics (GPR 4), were important for musicians as well as
nonmusicians, whereas other GPR's, such as the ones utilizing
articulation (GPR 1) and slur and rest (GPR 5), were much more
discernable to musicians than nonmusicians.
- When placed into conflict with other GPR's, GPR 7 (change of timbre)
was chosen quite frequently, indicating a high degree of salience for GPR 7.
Other GPR's, including the GPR for register (GPR 3) and the GPR for
attack-point (GPR 2) are also chosen quite frequently, although not as
frequently as GPR 7.
See also A
Generative Theory of Tonal Music.