Professional Documents
Culture Documents
INTRODUCTION
Singing is a complex process, using the respiratory, phonatory,
resonatory, articulatory, and auditory systems to create melodious vocal tones. In the general population, there is a significant
variation in singing ability. Individuals may be classified as accurate singers or inaccurate singers. To sing accurately, individuals must first be able to accurately hear, differentiate, store,
and then vocally reproduce pitches. Inaccurate singers, or
monotones, may have difficulty with these abilities.1 Thus,
pitch discrimination and pitch-matching tasks may be useful
in understanding factors that differentiate accurate singers
from those who are unable to sing accurately. Trained musicians and singers perform more accurately on pitch discrimination and pitch-matching tasks than most untrained individuals
(UT).24 Research suggests reliance on working memory during pitch discrimination tasks.57 Adding time delays between
tones,8 presenting tonal interference between the reference
tone and comparison tone,6,9 and using tones of differing timbre10,11 negatively affect pitch discrimination accuracy.
In addition to discriminating between two individual pitches,
individuals also determine differences between combinations
of pitches. Hubbard12 investigated the ability to discriminate
major triad chords in different positions (ie, tonic, first inversion, and others). A chord is the simultaneous sounding of
more than two notes at a time. A triad consists of three specific
notes sounded simultaneously. It is considered to be in root position when all of the intervals between the notes are in thirds,
Accepted for publication October 23, 2009.
From the Department of Speech Pathology and Audiology, College of Allied Health
Professions, University of South Alabama, Mobile, Alabama.
Address correspondence and reprint requests to Julie M. Estis, Ph.D., CCC-SLP, Department of Speech Pathology and Audiology, College of Allied Health Professions, HAHN
1119, 307 N. University Blvd., Mobile, AL 36688-0002. E-mail: jestis@usouthal.edu
Journal of Voice, Vol. 25, No. 2, pp. 173-180
0892-1997/$36.00
2011 The Voice Foundation
doi:10.1016/j.jvoice.2009.10.010
with the bottom note being the tonic note, or root, of the major
or minor scale on which it is based (eg, a C major chord consists
of the notes middle C, E, and G). Conversely, it is said to be an
inversion when the bottom note of the chord is either the third or
fifth of the triad (eg, the first inversion of a C major chord consists of the notes E [third], G [fifth], and C [tonic]). The stimulus
triads were simple major triads, so that people could participate
in the study, regardless of musical training. All participants
were considered untrained. Participants were asked to determine whether or not two sequentially presented target chords
were the same or different. Results of the study indicated that
listeners could accurately discriminate chords based on the
same root note. Hence, whether or not the chord was in root position, or inverted, the participants could tell if the chords were
the same, or different, harmonically. These results imply that
the ability to discriminate among the triads is based on cognitive processes rather than on perception of harmonics alone
or on previous musical training.
Most research investigating pitch processing and pitch
memory has focused on pitch discrimination tasks. Although
findings from pitch discrimination tasks may be implied for
pitch-matching abilities, it is necessary to systematically and
directly study pitch-matching abilities in a variety of populations
and with varied types of stimuli and interference to understand
the relationship between pitch processing and vocal production
of pitch. Thus, many aspects of pitch matching remain uncertain.
The known prerequisites for accurate pitch matching include
accurate pitch discrimination skills, normal auditory functioning, and good control of the vocal mechanism.11 Auditory feedback has also been shown to play an important part in accurate
pitch matching, particularly for trained and professional
singers.13 Much like pitch discrimination, pitch-matching skills
are also known to vary across populations (eg, TS vs UT).
Estis et al14 investigated pitch memory in a pitch-matching
task, specifically exploring the role of time delays on pitch-
174
matching ability. Results of the study indicated decreased pitchmatching accuracy with increasing time intervals of silence (5,
15, and 25 seconds) between the presentation of target tones and
vocal pitch-matching productions. Also, this study indicated
that some individuals with no formal vocal training performed
as well as vocally trained individuals, whereas a subset of UT
performed poorly on all pitch-matching tasks.
It remains to be determined which types of interference (eg,
time, vocal tones, pure tones, chords, and others) are most detrimental to pitch-matching performance. In addition, there has
been evidence to suggest a relationship between abilities in
pitch discrimination and pitch matching; however, research in
this area has resulted in mixed findings based on the populations
studied and the tasks used to measure pitch discrimination performance.3,6,11 Therefore, the purpose of the present investigation was to determine the effects of different types of auditory
interference between stimuli and vocal match productions on
pitch matching in males and females as well as to further investigate the relationship between pitch discrimination and pitchmatching abilities. This study also investigated the relationship
between vocal training and pitch-matching abilities with various types of interference, by including two groups of participants: trained singers (TS) and UT. The different types of
interference were musical chords of varying relation to the target tones and pink noise. If tonal information and speech information are held within specialized mechanisms in working
memory, then chords, particularly ones that are less related to
the target, should be detrimental to pitch memory and ultimately pitch matching. The following research questions
were addressed:
1. Are there differences in pitch-matching accuracy given
various types of interferences (musical chords with target
in root position, musical chords with the root a perfect
fourth away from the target, musical chords with the
root a major second away from the target, unrelated minor second chords an octave away, and pink noise) among
TS and UT?
2. Is there a relationship between the ability to discriminate
between pitches and the ability to match pitches?
METHODS
Participants
The participants were 20 females and 20 males between the
ages of 19 and 32 years (mean [M] 23.05, standard deviation
[SD] 3.08). All participants were native speakers of English;
had no significant history of voice pathology or voice treatment;
demonstrated adequate vocal function as evidenced by jitter
(frequency perturbation), shimmer (amplitude perturbation),
and noise-to-harmonic ratio (amount of noise in the signal)
within one SD of the mean, as calculated by Multi-Dimensional
Voice Profile-Advanced (MDVP-A) voicing analysis software
version 2.7.0 and compared with MDVP-A database means
and SDs; and passed an audiometric hearing screening. Participants were divided into two groups based on vocal training.
Twenty participants were TS. Participants in the TS group
had a minimum of 3 years of individual voice training and at
least 1 year of collegiate musical theory. The remaining 20 participants were UT who had no individual training from a professional vocal instructor.
Stimuli
For all pitch-matching tasks, stimuli were complex tones with
the following fundamental frequencies: 262 (C4), 294 (D4),
330 (E4), 348 (F4), and 392 Hz (G4) for female participants;
and 131 (C3), 147 (D3), 164 (E3), 175 (F3), and 196 Hz (G3)
for male participants. These frequencies were chosen because
they are within the normal singing range of females and males
in the third and fourth octaves of the musical scale. Tones and
chords were generated using a Yamaha Portable Grand keyboard (model YPG-625; Buena Park, CA) on a piano setting.
The stimuli were then recorded and saved onto a computer as
WAV files. Stimuli were edited with Adobe Audition (version
1.5) such that each tone had a duration of 1.5 seconds and
was gated on and off with 10-millisecond linear ramps.
Stimuli for the five interference conditions were created for
each target note (Table 1). For the five interference conditions,
there was a 1.5-second silence interval between the target tone
and the interference (chords or noise), which also had a duration
of 1.5 seconds. There were four chord interference conditions.
In the first condition (chord 1), the target tone was the root note
of the interfering chord, which was a major triad (eg, for the C3
and C4 targets, the interfering chord was C E G). In the second
condition (chord 2), the root of the interference chord was a perfect fourth above the target (eg, for the D3 and D4 targets, the
interfering chord was G B D). In the third condition (chord
3), the interfering chord had a root note that was a major second
above the target (eg, for the E3 and E4 targets, the interfering
chord was F] A] C]); and in the fourth condition (chord 4),
the interfering chord had a root note that was an interval of
an octave plus a minor second, or ninth, above the target tone
(eg, for the F3 and F4 targets, the interfering chord was be G[
B[ D[). It should be noted that the target note in its root position
was sounded in chord 1, at the interval of a perfect fifth in chord
2, and was not sounded in chords 3 and 4. The progression of the
four triadic conditions systematically increased intervallic distance or harmonic association from the target pitch. In addition
to four types of musical interference, an aperiodic noise condition was created. Specifically, pink noise with a spectrum similar to that of speech was generated with Adobe Audition sound
editing software (version 1.5).
For the pitch discrimination task, Adobe Audition sound editing software (version 1.5) was used to create complex tonal
stimuli. Five complex tones were created for the male and female participants based on the normal signing range for males
and females. The frequencies for the male participants were
104, 107, 110, 113, and 116 Hz. The frequency interval between complex tones was 50 cents. Each tone had a duration
of 1.5 seconds and was gated on and off with 10-millisecond
linear amplitude ramps. Each complex tone was paired with
each of the other complex tones and with an identical complex
tone for a total of 25 pairs of tones. This resulted in tone pairs
differing by 0, 50, 100, 150, or 200 cents. Each tone in a pair
was separated by a 0.5-second silent interval. Five complex
Julie M. Estis, et al
175
TABLE 1.
Complex Tones and Interfering Chords for Pitch-Matching Tasks
Males
Target Note
Coordinating
Frequency (Hz)
C3
131
D3
147
E3
164
F3
175
G3
196
Females
Chords
Target Note
Coordinating
Frequency (Hz)
1: C E G
2: F A C
3: D F] A
4: D[ F A[
1: D F] A
2: G B D
3: E G] B
4: E[ G B[
1: E G] B
2: A C] E
3: F] A] C]
4: F A C
1: F A C
2: B[ D F
3: G B D
4: G[ B[ D[
1: G B D
2: C E G
3: A C] E
4: A[ C E[
C4
262
D4
294
E4
330
F4
348
G4
392
tones were created for the female participants in the same manner at 200, 206, 212, 218, and 224 Hz. Tonal stimuli were presented via a Fostex 7301B3E (Tokyo, Japan) amplified speaker
at 75 dB sound pressure level (SPL) for all experimental tasks.
Volume settings were measured before the onset of the study to
ensure that all output was consistently at 75 dB SPL. In addition, sound level measurements were repeated after the study
to ensure that output level remained consistent.
Procedures
All procedures were conducted during a 1-hour session in a double-walled, sound-attenuated booth. Preexperimental tasks
were completed first. Participants read and signed a Statement
of Informed Consent. A bilateral pure tone hearing screening
was conducted using a Grason-Stadler, Inc (GSI-17; Milford,
NH) portable audiometer, calibrated in compliance with American National Standards Institute15 guidelines. Pure tones at
500, 1000, 2000, and 4000 Hz were presented at 25 dB HL
via TDH 50 (Telephonics, Farmingdale, NY) supra-aural headphones. Participants were instructed to raise their hand when
a tone was heard. Failure to respond at any frequency at either
ear precluded participation in the study.
Voice analysis was completed using the MDVP-A to ensure
that there were no current voice problems that may have adversely affected performance on pitch-matching tasks or accuracy of acoustic measurement of pitch-matching responses.
A head-mounted microphone was placed at approximately
34 cm from the left corner of each participants mouth for
recording responses. Participants sustained the /a/ sound at a
comfortable loudness level for 4 seconds. The subsequent vocal
Chords
Chord 1: C E G
Chord 2: F A C
Chord 3: D F] A
Chord 4: D[ F A[
Chord 1: D F] A
Chord 2: G B D
Chord 3: E G] B
Chord 4: E[ G B[
Chord 1: E G] B
Chord 2: A C] E
Chord 3: F] A] C]
Chord 4: F A C
Chord 1: F A C
Chord 2: B[ D F
Chord 3: G B D
Chord 4: G[ B[D[
Chord 1: G B D
Chord 2: C E G
Chord 3: A C] E
Chord 4: A[ C E[
176
FIGURE 1. Individual mean semitone difference scores in the immediate pitch matching condition.
condition as the within-subjects factor was performed. Results
showed a significant main effect for group (TS vs UT)
F(1,38) 17.008, P < 0.001, h2p 0.309with TS showing
significantly better pitch-matching accuracy than UT. There
was no significant main effect of pitch-matching condition
and no significant interaction.
Examination of individual performances as well as group
means and SDs revealed high variability in the UT, with some
individuals presenting very poor accuracy across all pitchmatching conditions. To further explore differences across
pitch-matching conditions, the UT was divided based on
pitch-matching accuracy in the immediate pitch-matching condition. Two groups of UT were created: UT-accurate (UT-A)
(10 participants with mean semitone difference scores less
than 1 semitone from the target tone; M 0.2377,
SD 0.2211) and UT-inaccurate (UT-I) (10 participants with
mean difference scores more than 1 semitone from the target
tone; M 3.1299, SD 1.2433). This division of participants
based on performance led to three groups with significantly
Julie M. Estis, et al
177
condition.
condition.
different pitch-matching accuracy in the immediate pitchmatching conditionF(2,37) 84.892, P < 0.001, h2p 0.821.
See Figure 7 for group means and SDs for semitone difference
scores across pitch-matching conditions. Because the UT-I
group performed poorly on pitch-matching tasks regardless of
interference, examination of the impact of interference on pitch
matching was compared between the TS group and the UT-A
group. Therefore, descriptive statistics and a 2 (group) 3 6
(condition) repeated-measures ANOVA with group as the between-subjects factor and pitch-matching condition as the
within-subjects factor were conducted. Mauchlys test of sphericity was significant, indicating that sphericity could not be assumed (w 0.032, P < 0.001); therefore, Huynh-Feldt
corrections were used. A significant main effect for pitchmatching condition was shownF(2.608,73.036) 4.255,
P 0.011, h2p 0.132. The accuracy of the TS diminished
with several conditions; however, their performance was more
consistent and accurate than that of the UT-A group. Analysis
of between-subject effects yielded a significant main effect
for groupF(1,28) 5.651, P 0.016, h2p 0.189. The interaction was nonsignificant. Post hoc pairwise comparisons revealed that chord 2 (P 0.007), chord 3 (P < 0.001), chord 4
(P < 0.001), and noise (P 0.007) interference conditions
were significantly different from performance in the immediate
condition. Also, there is a significant difference in performance
between the chord 3 and chord 4 conditions and between the
chord 3 and the noise conditions.
condition.
condition.
178
TABLE 2.
Group M and SD of Semitone Difference Scores for TS and UT
Trained Singers
Pitch-Matching Condition
Immediate
Chord 1
Chord 2
Chord 3
Chord 4
Noise
Untrained Individuals
SD
SD
0.1246
0.1601
0.1645
0.2539
0.2830
0.1835
0.0567
0.1283
0.1115
0.2340
0.1979
0.1324
1.6838
2.0480
2.0450
2.1740
2.0694
2.0047
1.7195
2.6246
2.0683
1.9357
1.9757
2.1132
UT-I groups.
tion task.
Julie M. Estis, et al
However, chords that did not contain the target tone (chord 3
and chord 4 conditions) negatively impacted pitch-matching accuracy. As anticipated, results showed increasingly significant
differences between the immediate condition and chord 2,
chord 3, and chord 4 conditions. The chord 3 (musical chords
with the root a major second away from the target) and chord
4 (unrelated chord a ninth above) conditions were found to be
most detrimental to the untrained accurate singers when compared with their scores in the immediate condition. Noise interference also significantly reduced pitch-matching accuracy. In
summary, the current study reveals that TS have pitch-matching
skills superior to those of UT. However, some UT demonstrate
pitch-matching accuracy similar to TS. This indicates that, in
addition to musical exposure and learning, innate factors may
play a role in pitch-matching abilities. The consistently accurate performance observed in TS suggests that vocal training
fine-tunes the underlying mechanisms involved in pitch matching, thereby enhancing pitch-matching accuracy. Specifically,
practice and training likely improve the efficiency of the vocal
mechanism, allowing the TS to quickly and precisely configure
the vocal folds for production of a specific pitch. Additionally,
musical training yields improved cognitive representations for
musical notes, enhancing the efficiency of the perceptual and
memory resources for pitch.
179
SUMMARY
Musical training enhances pitch-matching accuracy. For TS,
pitch-matching ability remains strong despite musical and noise
interference. Pitch-matching accuracy varies considerably
among UT. Those who show accurate pitch matching without
a musical background are more susceptible than TS to reduced
pitch-matching accuracy when chords that do not contain the
target tone and noise interference are presented. Those who
are unable to adequately match pitch show poor performance
with and without musical and noise interference. This study
also supports previous research showing a strong relationship
between pitch discrimination and pitch-matching abilities.
This study provides insight into the underlying processes involved in pitch matching and pitch discrimination. Results suggest that pitch memory is enhanced by musical training,
although some individuals without training appear to show naturally strong pitch memory skills. Questions remain regarding
the exact cognitive, perceptual, and physiological mechanisms
involved in pitch matching. Future research exploring neurological, auditory, and physiological correlates may explain the
variation in pitch matching among the general population and
the effects of vocal training on these underlying systems.
Results of this study imply that individuals who demonstrate
poor pitch matching and poor pitch discrimination may require
additional training. For example, vocal function exercises,20
which require patients to sustain sounds at various frequencies,
are used by speech language pathologists to treat a variety of
vocal pathologies. Some patients may demonstrate difficulty
with these tasks because of reduced pitch perception and
pitch-matching abilities. These patients may require modifications to the typical treatment protocol. Also, choral directors
and teachers of singing may consider incorporating similar
pitch matching and pitch discrimination tasks into their auditioning process to screen UT. This would allow for determination of those who show naturally strong pitch-matching and
pitch discrimination abilities.
REFERENCES
1. Watts C, Murphy J, Barnes-Burroughs K. Pitch matching accuracy of
trained singers, untrained subjects with talented singing voices, and untrained subjects with nontalented singing voices in conditions of varying
feedback. J Voice. 2003;17:185194.
2. Fujioka T, Trainor L, Ross B, Kakigi R, Pantev C. Musical training
enhances automatic encoding of melodic contour and interval structure.
J Cogn Neurosci. 2004;16:10101021.
3. Murry T, Zwirner P. Pitch matching ability of experienced and inexperienced singers. J Voice. 1991;5:197202.
4. Schueller M, Bond Z, Fucci D, Gunderson F, Vaz P. Possible influence of
linguistic and musical background on perceptual pitch-matching tasks: a pilot study. Percept Mot Skills. 2004;99:421428.
5. Deutsch D. The deterioration of pitch information in memory [unpublished
doctoral dissertation]. San Diego, CA: University of California at San
Diego; 1970.
6. Moore R, Keaton C, Watts C. The role of pitch memory in pitch discrimination and pitch matching. J Voice. 2007;21:560567.
7. Pechmann T, Mohr G. Interference in memory for tonal pitch: implications
for a working-memory model. Mem Cogn. 1992;20:314320.
8. Harris J. The decline of pitch discrimination with time. J Exp Psychol.
1952;43:9699.
180
9. Deutsch D. Tones and numbers: specificity of interference in short-term
memory. Science. 1970;168:16041605.
10. Erickson M, Perry S. Can listeners hear who is singing? A comparison
of three-note and six-note discrimination tasks. J Voice. 2003;17:353369.
11. Watts C, Moore R, McCaghren K. The relationship between vocal pitchmatching skills and pitch discrimination skills in untrained accurate and inaccurate singers. J Voice. 2005;19:534543.
12. Hubbard T. Listeners can discriminate among major chord positions. Percept Mot Skills. 1998;87:891897.
13. Murbe D, Pabst F, Hofmann G, Sundberg J. Significance of auditory and
kinesthetic feedback to singers pitch control. J Voice. 2002;16:4451.
14. Estis J, Coblentz J, Moore R. Effects of increasing time delays on pitchmatching accuracy in trained singers and untrained individuals. J Voice.
2009;23:439445.