Information Technology Reference
In-Depth Information
Table 4. Results of pitch interval and pitch contour
performance in Experiment 3. Lower two rows are
comparative results from Lindsay (1996)
Subject
Table 5. Results of pitch interval reproduction
performance in Experiment 3, separated into trials
common with Lindsay (1996) and modified trials
Subject
Pitch Interval Matching
Contour
Common Trials
Modified Trials
Group
1
st
1
st
/2
nd
Matching
Group
1
st
1
st
/2
nd
1
st
1
st
/2
nd
MUSI
65.2%
77.2%
81.1%
MUSI
76.2%
84.5%
52.7%
68.9%
BTWN
48.0%
64.0%
72.0%
BTWN
64.7%
75.0%
28.1%
50.9%
NONM
47.1%
65.5%
69.7%
NONM
56.7%
73.3%
37.3%
57.6%
ALL
54.5%
69.7%
74.9%
ALL
67.0%
78.3%
40.5%
60.0%
LIN-M
86.3%
96.9%
95.6%
Lindsay
Common Trials
6-semitone trials
LIN-N
40.6%
68.8%
71.9%
LIN-M
90.6%
94.1%
81.3%
94.7%
LIN-N
41.2%
64.7%
40.0%
73.3%
Figure 2. Performed intervals vs. stimulus intervals
for MUSI subjects, all 32 trials, in Experiment 3
measuring by pitch intervals, and 71% matched
the intended pitch contour. The values for each
group are summarized in Table 4.
To better compare results between the two
studies, it is helpful to break down the data into
two groups: results based on the 17 unmodified
note sequences which were common to both
Lindsay's study and our own, and those based
on the 15 modified phrases. When we consider
these two components separately, we see clearly
the effect of the larger intervals tested in our
experiment, as shown in Table 5. In the unmodi-
fied trials, our MUSI group matched the correct
phrase using pitch interval data 76% of the time,
while Lindsay's musicians scored 91%; the dif-
ference between the groups is smaller than the
20% overall difference when including all 32
trials. The modified set of trials posed more of a
difficulty to our subjects, as performance dropped
dramatically to 53%; in comparison, Lindsay's
musician group showed a significant but much less
dramatic decrease to 81% for the trials involving
six-semitone intervals. The BTWN and NONM
groups showed the same behavior as the MUSI
group. Interestingly, Lindsay's nonmusician was
the only one to score equally well between the
two subcategories.
Figure 2 illustrates the collective values of
pitch intervals hummed by the MUSI group, the
best performers of the three groups reported. This
plot is helpful to illustrate instances where pitch
contour errors occurred. Any points which ap-
pear in quadrants II or IV of the graph represent
intervals which were in the opposite direction
of the associated stimulus interval. It can be
seen from the graph that contour errors occur
against both large and small stimulus intervals,
and they are distributed evenly between upward
and downward intervals. Furthermore, when a
contour mistake is made, the sizes of the incor-
rectly hummed intervals in these cases also vary
Search WWH ::
Custom Search