Materials and Procedure
The stimuli consisted of the 15 successive
solid or arpeggiated chords that constitute the first nine bars of the
second movement of the Waldstein. The sonata was performed by an accomplished
pianist and recorded in MIDI format. Chords were then extracted from the
MIDI file. The chords were heard through Sennheiser HD-480 headphones connected
to a Roland FP-1 digital piano. A Macintosh computer running MAX software
controlled the timing and presentation of stimuli and the collection of
responses. After each presentation of a chord, listeners rated perceived
dissonance on a six-point scale. Each of eight blocks of trials contained
each of the 15 chords, randomly ordered. The first block was a practice
block, and its data were not included in the analyses. In order to avoid
any carryover effects of the tonality of the previous trials the chord
heard on each trial was randomly transposed within the range 3 semitones
below to 3 semitones above the original notation of the excerpt.
Results
Perceived dissonance varied significantly among
different chords (F(14,182) =26.70, p <. 001). The mean dissonance rating
for each chord is plotted in Figure 1, where the x-axis represents the
successive time points, from 1 to 15, at which each chord was extracted.
Participants
Fourteen volunteers from the Queen's University
community all met the same musical training requirement as Experiment
1, and had an average 12.3 years of musical training. They had all participated
in a previous experiment in which they had been required to perform the
Waldstein excerpt from memory.
Materials and Procedure
The stimuli consisted of 15 segments from the Waldstein.
Each segment contained all the musical material up to one of 15 successive
time points in the first nine bars of the second movement. Thus the
final chord of each segment was one of the chords tested in Experiment
1. The segments were presented in chronological order and were not randomly
transposed between trials. In other respects, the procedures was
the same as Experiment 1. After each presentation listeners rated the perceived
tension at that time point on a six-point scale.
Results
Listeners' ratings of perceived tension varied across
time points (F(14,182) = 37.91, p < .001). The mean tension rating for
each segment is plotted in Figure 1, where the x-axis represents the successive
time points at which each segment stopped. It can be seen that the fluctuation
in tension ratings for the segments closely resembles the fluctuation in
dissonance ratings for the isolated chords. As well as the fluctuations,
however, the tension ratings show an increase up to time point 8
that is not present in the dissonance ratings. This difference may be attributed
to the phrase structure of the excerpt.
The tension ratings collected in the Experiment 2 were highly
correlated (r(13) = .90, p < .001) with the dissonance ratings collected
in Experiment 1. In addition, however, the tension ratings were also highly
correlated (r(13) = .65, p < .01) with a quantified predictor
representing phrase structure. This predictor was derived from a music-theoretic
analysis of the excerpt (Lerdahl, 1988). A multiple regression was then
performed using dissonance ratings and phrase structure as predictors of
perceived tension (R = .96). A significant contribution to the predictability
of perceived tension was made by both the dissonance ratings (t = 8.67,
df = 12, p < .001) and phrase-structure (t = 4.23, df = 12, p
< .01).

Lerdahl, F. (1988). Tonal pitch space. Music Perception,
5, 315-350.