Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

SIGHT-SINGING PERFORMANCE AND PIANO ACCOMPANIMENT

Philip Fine & Hazel Younger


University of Buckingham, UK & St. Hilda’s College, University of Oxford, UK.

Expectation is important in music reading, as demonstrated in


ABSTRACT priming studies and studies of proof-reader’s error in pianists.
Sight-reading improves when the piece conforms to a familiar style,
Previous research into sight-singing suggests that hearing the as this makes expectation more accurate. The intonation of most
accompaniment can improve pitching accuracy (Fine, Berry & instrumentalists improves with accompaniment, and the same could
Rosner, in preparation). This study additionally investigated the be expected of sight-singing performance. Pitching and intonation
effect of seeing the accompaniment. Using an objective measure of inaccuracy in singers can have two causes; incorrect intervals or
accuracy and a within-subjects design, 48 singers were tested, but pitch drift. Incorrect intervals are caused by misrepresentation or
no significant effect of the presence of visual or aural miscalculation of the pitch that is being attempted. Pitch drift
accompaniment was found. Significant interactions showed that causes inaccuracy by a gradual change in the representation of the
aural accompaniment increased improvement between attempts (F key note (often becoming slightly flat), or each interval being
(1, 42) = 4.51, p < 0.04), and that the skill of the singer affected the slightly too small or large, depending on whether it is ascending or
use of visual accompaniment (F (1, 21) = 6.20, p < 0.03). The findings descending.
have both educational and practical implications.
Singing performance of familiar melodies is best with tonal
1. INTRODUCTION accompaniment, as opposed to chromatic or dissonant. Intervals are
also sung more accurately when presented in context. Boyle and
Sight-reading is performing an unfamiliar piece of music with no Lucas (1990) found that singing performance was better with
prior practice. It is an important skill for professional musicians and accompaniment than without, especially in less experienced
involves transforming a visual input (the score) into a motor output, students, but used a simplistic scoring system with no allowance for
such as pressing a piano key. Translation of the visual input a correct interval sung from a preceding incorrect note. Fine et al
involves recognizing groups of notes and reading ahead. (in prep.) used a more complex scoring system, investigating the
Most sight-reading research has focused on pianists. However, effect of accompaniment on sight-singing accuracy, and separate
despite the differences between instrumental sight-reading and scores were produced for the number of correct notes and the
sight-singing, there has been little research into sight-singing. This number of correct intervals. Predictability of both harmony and
is perhaps surprising, considering its practical importance to melody significantly improved pitching accuracy, less skilled
singers and the benefits it can bring to other musicians, such as readers being more affected than more skilled readers, supporting
improved instrumental sight-reading. the previous findings. They also showed that a second attempt was
significantly more successful than the first, which has practical
In singing, the motor output cannot generally be represented as implications in terms of the fairness of sight-reading tests in
explicitly as in instrumentalists. Singers need to know the sound of auditions and examinations.
a note before producing it, i.e. they must internalize the pitch.
Although such auditory representations have been shown to be On occasion, singers will change the pitch of a note once started.
involved in piano playing (Waters, Townsend & Underwood, 1998), This self-correction presumably occurs when, after producing a
many instrumentalists do not have to hear the pitch of the note note, the singer realizes that it is incorrect. This is probably affected
before producing it (e.g. pianists just have to press the right key). by accompaniment, as dissonance can indicate error.

Singers, however, need a starting note before they can perform Seeing the accompaniment may also affect sight-singing, as other
(apart from the few with perfect pitch). Then, instead of linking parts in the score provide useful pitching cues (Fine, 2002).
each note to a motor representation of its performance, a singer will Experienced musicians can identify chordal structure, which might
analyze the interval from the preceding note(s) (Fine, 2002), and increase their ability to judge intervals more accurately, even in the
change pitch appropriately. absence of hearing the accompaniment. Seeing the accompaniment
may also reduce self-correction, as knowing how the melody fits in
A further difference between singing and instrumental performance with the piano part may lessen the tendency to change pitch after
is that singers rarely perform alone and unaccompanied. This production.
presents the possibility that the accompanying instruments or
singers provide cues for pitch (Fine, 2002; Fine et al, in prep.). Although Boyle and Lucas (1990) and Fine et al (in prep.) both
showed that the presence of accompaniment improves
Accompaniment could aid singers in two ways: by assisting sight-singing performance, scoring in both cases was done
intonation, and by helping them make judgments about intervals. A subjectively within arguably flawed scoring systems. As pitch
tonal accompaniment, for instance, could limit the selection of perception is generally categorical, subjective scoring could be
plausible choices through harmonic and melodic expectation. biased. For example, if a singer sings a slightly reduced ascending

ISBN 1-876346-50-7 © 2004 ICMPC 778


ICMPC8, Evanston, IL, USA August 3-7, 2004

interval, the pitch of the note may become technically wrong but Procedure. Participants were tested singly. They were given about
the listener’s expectation can lead them to hear it as correct but flat. 15 seconds to look at each piece before attempting it (treble or bass
clef according to preference), and a further 10 seconds between
The main aim of this study was thus to assess the effects of aural attempts. A metronome beat was provided to keep participants in
and visual presentation of the accompaniment on pitching accuracy time (crotchet = 100). All trials began with presentation of the key
during sight-singing, using an objective measure of pitch. chord and the starting note, and at least a bar of metronome beats.
Participants sang each piece twice. It was expected that singers Accompanied attempts began after a bar of metronome and any
would perform better on the second reading of a piece, and that both piano introduction, and unaccompanied after a bar of metronome,
hearing or seeing the accompaniment would improve pitching but with the first three notes doubled on piano, to ensure the
ability, though the effect of seeing the accompaniment may be participants began at the right time, and to improve confidence in
limited to more experienced musicians. It was further expected that the starting note and key.
self-correction would be more common when the piano
The importance of keeping going and not backtracking to correct
accompaniment could be heard and when it could not be seen.
errors was emphasized. Occasionally an attempt was abandoned
Finally, few studies have looked at the opinions of sight-singers during the first few bars due to confusion over tempo or
regarding preference for being accompanied. Anecdotal evidence introduction, in which case participants were allowed an extra
suggests that singers prefer to hear the accompaniment, and prefer attempt; otherwise scoring was as normal if an attempt was not
to see the accompaniment when it can be heard, but that they are completed.
indifferent to visual presentation alone. This was investigated. The pitch of each note produced was then identified using Praat.
Pitches within a semitone (50 cents) of the required note were
2. METHOD accepted as correct. The number of correct pitches was then
Participants. Twenty males and 28 females (mean age 20, ranging counted and turned into a percentage score for each piece. As
from 15 to 24) participated, recruited from various choirs including intervals depend on pitch ratios, it was decided that an acceptable
the National Youth Choir of Scotland. All participants had some interval would fall within the size of [note 1 - note 2 r half a
experience of sight-singing, usually in examinations or auditions, semitone]. Self-correction was assessed subjectively with the help
although some were primarily instrumentalists rather than singers. of Praat.
Those with self-reported perfect pitch were excluded, and all were
fluent readers of both treble and bass clef. Allocation to condition 3. RESULTS
was random. Age, sight-singing experience and preference for
seeing and hearing the piano part (on a three point scale: prefer, As the pitch and interval scores were in the form of percentages, an
indifferent, prefer not) were recorded. Arcsine transformation was performed on the data. A repeated
measures ANOVA, with the within-subjects factors of piece (4
Design. There were 3 dependent variables: the percentage of levels) and attempt (2 levels), showed a significant effect of attempt,
correct notes, the percentage of correct intervals, and the number of with improvement on the second attempt, on both pitch (F (1, 41) =
occurrences of self-correction. There were 2 independent variables 5.14, p < 0.03) and interval (F (1, 41) = 19.90, p < 0.001) scores, and
related to accompaniment: seeing and hearing. Combination of also of piece on pitch (F (3, 123) = 6.49, p < 0.001) and interval (F (3,
these within-subject variables resulted in four possible conditions. 123) = 34.83, p < 0.001) scores. The interaction was not significant.
These were presented to the participants, each condition paired Piece A led to fewest errors, D most, probably due to differences in
with one of the four test pieces. The third within-subjects the interval distributions.
independent variable was attempt, as participants were asked to
sing each test piece twice. Order of conditions and pieces was The same independent variables were used to test for an effect on
counterbalanced. the number of self-corrections. Attempt but not piece was found to
be significant, with fewer self-corrections in the second attempt (F
Materials. The test pieces were selected from published examples
(1, 41) = 12.40, p < 0.001). The mean scores for pitch, interval and
of ABRSM (Associated Board of the Royal Schools of Music)
self-correction are shown in Table 1.
grade 7 singing standard sight-singing tests. Selection was based on
rhythmic simplicity (as complex rhythms could distract participants) Performanc Pitch % Interval % No. of Self
and vocal range. The pieces were rewritten using Coda Music e Corrections
Finale Notepad software to present to participants. The number of Score
metronome beats preceding the piece was also shown on the score.
Piece A 62.68 76.15 0.49
Accompaniment, played on a piano and not synthesized, was Piece B 53.24 55.93 0.40
recorded on a Sony minidisc player and transferred to CD. The Piece C 57.91 62.49 0.37
metronome beat was recorded simultaneously. The accompaniment Piece D 51.62 57.01 0.36
was presented using a portable Sony CD player through earphones, Mean 56.55 63.10 0.42
at a preset volume – participants were told that they could adjust it Table 1: Mean performance scores by piece
if they wished, but none did. Attempts were recorded onto Sony
minidisc and analyzed using the freeware computer program Praat, A second within-subjects ANOVA with the two accompaniment
which was used to identify the fundamental frequency of each note. independent variables (seeing and hearing) and attempt was carried

779
ICMPC8, Evanston, IL, USA August 3-7, 2004

out on the pitch and interval scores. Tables 2 and 3 display the Yes Don’t mind No
means by condition and attempt for pitch and interval scores Prefer hearing? 46 2 0
respectively. Neither seeing nor hearing the accompaniment had a
Prefer seeing when hearing? 24 23 1
significant effect on pitching accuracy. There was, however, a
Prefer seeing when not 6 42 0
significant interaction between hearing the accompaniment and
hearing?
attempt for pitch score (F (1, 42) = 4.51, p < 0.04). There was greater
Table 4: Response frequencies to preference questions
improvement when the accompaniment could be heard.

Accompaniment Condition Pitch score (%)


4. DISCUSSION
1 2 Mean The hypothesis that pitching accuracy would be greater on the
See / Hear 56.45 61.64 59.11 second attempt at a particular piece was supported, as found by
Hear only 50.56 60.01 55.56 Fine et al (in prep.). The lack of support for the hypothesis that
See only 55.55 53.55 54.53 hearing the accompaniment would improve pitching accuracy is
Neither see nor hear 56.68 58.33 57.47 surprising, given the results of previous studies (e.g. Boyle & Lucas,
Table 2: Mean pitch scores by condition and attempt 1990, Fine et al, in prep.). However, the latter study compared the
effect of harmonious accompaniment to chromatic or dissonant
accompaniment, suggesting that unpredictable accompaniment
Accompaniment Condition Interval score (%) may impair accuracy, rather than predictable accompaniment
1 2 Mean aiding it. The hypothesis stating that visual presentation of the
See / Hear 60.18 64.52 62.39 accompaniment would improve pitching accuracy was also not
Hear only 60.21 66.42 63.45 supported, suggesting perhaps that the singers were not paying
See only 61.75 62.17 62.05 attention to parts other than their own.
Neither see nor hear 62.87 65.69 63.17
Table 3: Mean interval scores by condition and attempt When participants heard the accompaniment, they improved more
between attempts than when they sang alone (as shown by the
There was a non-significant trend in the direction of increased significant interaction between hearing the piano and attempt).
self-correction when accompaniment could be heard, irrespective Being able to predict the piano part (through memory of the first
of whether or not it could be seen (F (1, 42) = 2.973, p < 0.09), and time) could help to predict the vocal line, or it may be that
when it was not seen, whether or not it could be heard (F (1, 42) = participants ignore the piano on their first attempt, and only after
3.31, p < 0.08). There was a significant interaction between hearing they are familiar with the melody do they begin to integrate their
the accompaniment and attempt on self-correction scores (F (1, 42) = knowledge of the accompaniment into their pitching tactics.
5.84, p < 0.02): a reduction in self-correction occurred only when
the accompaniment could be heard. Self-correction, when a singer changes the note they are singing,
was not significantly affected by either aural or visual
To investigate any effects of sight-singing skill, participants were accompaniment, although non-significant trends suggested that
split into 4 groups on the basis of their average pitch score. When self-correction may be encouraged both by hearing the piano and
the analysis was restricted to the top and bottom skill groups, there by not seeing the accompaniment. Self-correction occurred
was a significant interaction between seeing the accompaniment significantly less in the second attempt, although this may be due to
and skill level on interval scores (F (1, 21) = 4.54, p < 0.05), which the overall increase in scores – singers do not need to change from
was supported by a trend when all levels were included. There was notes that are correct, so are more likely to do so when they are
also a significant three-way interaction between seeing the performing poorly. The significant effect of attempt may also have
accompaniment, skill level and attempt on pitch scores (F (1, 21) = been caused by increased confidence in the pitching decisions made,
6.20, p < 0.03). The best sight-singers were better and improved or anticipation of those notes that were self-corrected during the
more when they could see the accompaniment, but the least first attempt. The interaction with hearing the accompaniment
experienced were better and improved more when they could not showed that this effect was larger when participants could hear the
see it. piano, suggesting that hearing the accompaniment in the first
attempt improved anticipation of potential mistakes in the second.
Participants’ response to each of the preference questions (coded as
1, 0 or -1) are shown in table 4. As no overall preference would As Fine et al (in prep.) found that sight-singing skill influenced the
result in an average value of 0, this was compared to each mean effect of the accompaniment, the singers were divided into four
preference score using a one-sample t-test. There was a significant equally sized groups, based on their average pitch score.
preference for hearing the accompaniment (t (47) = 32.879, p < Restricting the analysis to the highest and lowest scoring groups
0.001), for seeing the accompaniment when it was heard (t (47) = resulted in a significant three-way interaction between skill level,
6.087, p < 0.001) and for seeing even when the accompaniment was attempt and seeing the piano accompaniment. Only the most skilled
not presented aurally (t (47) = 2.591, p < 0.05). sight-singers improved when the piano part was visible, suggesting
that they were more adept at integrating their knowledge of the
piano part with the melodic line. This could be due to their relative

780
ICMPC8, Evanston, IL, USA August 3-7, 2004

ease of reading the tune allowing them the capacity to analyze the The significant effects and interactions on self-correction scores are
piano part, or it could be that more of them were pianists or more difficult to interpret. Self-correction demonstrates that the
instrumentalists and therefore could read piano parts more fluently. singer is aware of an error, but singers who are over-enthusiastic in
It would have been advantageous to ask participants whether they changing to fit with the piano part may run into difficulties when
could play the piano, which should be taken into account in future faced with an intended dissonance or suspension. It may be the case
studies. The least skilled singers were disadvantaged by seeing the that singers do not self-correct if the style of the music suggests that
piano part, perhaps indicating that attending to or attempting to the accompaniment is likely to be dissonant, which should be taken
analyze the accompaniment distracted from the melodic line. into account by future studies.

Most of the participants, however, reported no strong preference for Although the pieces were all ABRSM Grade 7 sight-reading tests,
visual presentation alone, and approximately half preferred to see they were found to differ in difficulty level as both pitch and
the piano part if it was being played. The preference for aural interval scores varied systematically between them. This is
presentation of the accompaniment was almost unanimous. These probably due to differences in rhythmic, structural and key
findings are surprising in the light of the lack of a main effect of complexity, not just pitching. For example, piece A, which had the
seeing or hearing the accompaniment, but suggest either that highest average score, contained many repeated notes, but was in a
accompaniment can be beneficial to singers in specific conditions difficult key and contained reverse dotted rhythms, which were
(for instance, if sight-singing in a choir, where other singers may often sung incorrectly. In comparison, piece D was rhythmically
make distracting errors, or for particular styles of music), or that it simple, but contained larger intervals, fewer repeated notes and
has benefits to other aspects of performance, such as increasing little use of the tonic triad. To tease apart the effects of
confidence or noticing errors. Hearing the accompaniment may accompaniment, which may be somewhat piece-specific, or at least
also improve rhythmic confidence, or simply increase the aesthetic reliant on the stylistic quality of the piano part, future studies
value of the music. Increasing confidence is in itself a desirable should use more equally matched pieces in terms of their style,
outcome, particularly as sight-singing can be a stressful experience, pitching and rhythmic difficulty.
no doubt worse in an exam or audition situation. Increased
confidence could promote louder singing, rhythmic accuracy and As the limited existing research supports the influence of an
expressive aspects of performance. accompaniment on pitching performance in sight-singing, future
research should perhaps test more restricted hypotheses or include
In this study, the scoring was objective, using Praat. It is possible qualitative research, because until it is understood how
that, had the pieces been scored subjectively, slightly different accompaniment exerts its influence, the implications of such
results might have been found, as subjective scoring is less research are necessarily vague. Future work should compare the
sensitive to pitch drift than objective, but on the other hand notes effect of piano accompaniment, which is likely to be correct, to the
out by as much as 49 cents were counted as correct, so intonation effect of sight-singing surrounded by other singers, who may be
was ignored. Obviously, in the real world of musicians, it is themselves making mistakes. It is clear that singers need to be
subjective and not objective evaluation that matters and this should taught to sight-sing in a more systematic manner – research into
therefore be combined with objective measurements in future method and intervention programmes would be beneficial – as the
research. overall standard was surprisingly low. Education should also
include instruction on how to make best use of accompaniment,
Comparing the results of this study to those of Fine et al (in prep.), especially when it is presented visually, as it seems likely that
the negative effect of dissonant or unpredictable accompaniment pianists have an advantage in this sense. Future study should look at
appears greater than the positive effect of tonal accompaniment. It methods of improving sight-singing so that these findings can be
might therefore be useful to learn modern pieces unaccompanied implemented in a practical sense.
(i.e. those that do not fit neatly with the piano part), but that other
pieces would not suffer from initial practice either with or without
the piano. The accompaniment may have had a significant effect 5. REFERENCES
had the sample been of a more limited skill range, or different
pieces used. However, while the results do not prove that 1. Boyle, J.D. & Lucas, K.V. “The effect of context on
accompaniment cannot affect pitching performance, they do show sightsinging.” Bulletin of the Council for Research in
that accompaniment does not necessarily affect pitching accuracy. Music Education, 106, 1-9, 1990.

Aural accompaniment was found to increase improvement between 2. Fine, P.A. “Note-finding strategies in singing: An
attempts, suggesting that accompanied tests of sight-singing would interview study on Schnittke’s Bussvers XII”. ICMPC 7.
be more representative of pitching ability if a second attempt was Sydney, Australia, 2002.
allowed. Skilled sight-singers were found to improve more when 3. Fine, P.A., Berry A. & Rosner, B. “The Effect of Pattern
accompaniment was presented visually, arguing against the use of Recognition and Tonal Predictability on Sight-Singing
part-books in choirs. Less skilled readers, however, may be Ability.” In preparation.
distracted by seeing the accompaniment, suggesting either that in
the early stages, single line music may be more appropriate, or that 4. Waters, A.J., Townsend, E. & Underwood, G. “Expertise in
novice singers should be taught how to make best use of the piano musical sight-reading: A study of pianists”. British Journal
accompaniment. of Psychology, 89, 123-149, 1998.

781

You might also like