An Electromyographic Analysis of Bowing Technique

Dr. William K. Koehler

Professor of Double Bass/Music Education

Illinois State University

    Musical performance on a stringed instrument is a complex process which involves cognitive as well as motor learning. String teachers have been successful in developing methodologies which foster virtuostic performance technique as result of increased emphasis on how various aspects of performance (bowing, shifting, vibrato finger motion, etc.) are executed. The traditional teaching approach typically consists of modeling by the teacher combined with physiological descriptions of the motions involved. While this approach often accomplishes its goal, other methods like those incorporating modern technological advances can be employed to enhance physiological understanding and modify the intricate motions required in string performance. Little is known about physiological and biomechanical aspects of the interaction of muscles in performance of various bowing types. Further, little is known about the extent to which feedback concerning muscle tension might affect performance quality. It follows that knowledge of the relationships among muscle groups involved in the performance of the various bowing types is of great interest to string performers and teachers.

      Biofeedback is one such means of providing objective information related to muscle  expenditure. It directly addresses efficiency of movement, may help to modify undesirable aspects in string performance, and can provide a means of examining anatomical aspects of music performance. Electromyographic biofeedback training has been successfully employed to reduce muscle tension levels and improve musical performance. Morasky, Reynolds, and Sowell (1983) used EMG biofeedback training to control excess muscle tension in clarinetists. Results revealed significantly lower within-session EMG levels on trills and scales between pretest to posttest. It was found that as EMG levels decreased, performance dexterity increased. Levee, Cohen, and Rickles (1976) conducted a case study employing EMG biofeedback for the relief of tension in the facial and throat muscles of a woodwind player. Increased musical proficiency was reported.

       In a case study by Le Vine (1983), a professional violist who experienced a craft palsy accompanied by obsessions and compulsions, was treated with in-vivo temperature biofeedback exercise. Improvements became evident, and improvements were in evidence four years after termination of treatment. Electromyography has been successfully employed as a means of measuring the effect of muscle tension on performance quality. In a study using electromyography to measure muscle tension, Dennis (1984) examined the effect on performance quality of three methods of supporting the double bass. Results indicated non-significant differences in either performance quality or muscle tension regardless of method of support. Cutietta (1986) examined the effects of biofeedback training on musicians while performing problem passages of music. Instructors' ratings revealed significantly better performances by the experimental group, and significantly lower EMG scores for all experimental group subjects with the exception of the percussionist.

      EMG biofeedback was employed as a means of reducing left-hand muscle tension of nine string players in a study by Morasky, Reynolds, and Clarke (1981). Decreases in EMG were significant on a number of within-session and intra-session conditions.  However, this design does not take into account differences in EMG potentials between the first and final sessions. In a study addressing pedagogical implications of biofeedback, White and Basmajian (1974) investigated muscle expenditure on facial muscles as a function of register, dynamic level, and proficiency level among trumpeters. Results revealed differing muscle activity preceding and following the production of tones between beginning and advanced trumpeters. Advanced trumpet players show a smaller ratio of upper to lower lip activity than beginning players. A reduction in the ratio between upper and lower lips may result in improved performance.

      In an exploratory study involving ten high school and university string players, Koehler (1989) employed EMG biofeedback training as a means of analyzing and improving bowing ability. While this exploratory study involved a rather small number of subjects, results indicated that subjects receiving biofeedback achieved significant reductions in EMG potentials on specific muscle groups of the bow arm. Results suggested that knowledge of the correlation between posterior deltoid and forearm flexor muscle groups may enhance performance ability on spiccato and sautillé bowing strokes.

      The purpose of this study was to gather data on muscle expenditure and to examine the effect of Biofeedback training on string players' performance of spiccato and sautillé bowings. This study employed EMG biofeedback training to reduce muscle tension in four muscle groups of the bow arm. The study assessed string performance achievement on five dependent measures: (1) Rhythmic Accuracy, (2) Articulation, (3) Tone Quality, (4) Tempo Consistency, and (5) Intonation. Specific research questions included:

To what extent are muscle expenditure data intercorrelated?

What are the effects of treatment conditions on performance of spiccato and sautillé bowing tasks?

In relation to muscle expenditure data, to what extent does performance of spiccato and sautillé differ by instrument and the interaction of instrument by treatment?

To what extent do treatment conditions affect muscle expenditure on spiccato and sautillé bowing tasks?

To what extent are musical performance variables related on measures of spiccato and sautillé bowings.

To what extent are musical performance variables intercorrelated?

    The sample was comprised of 47 string players. All subjects had at least one year of high school orchestral experience. Subgroups included: 8 high school, 31 university undergraduate, 4 university graduate, and 4 post-university string players. Post university string players were defined as those who either were retired university professors, orchestral players completing at least a masters degree in music, or those with extensive music experience. Subjects were classified by instrument (12 violin, 9 viola, 8 cello, 18 double bass) and experience level. Subjects within each instrumental group were randomly assigned to experimental (n = 24) and control groups (n = 23).  A Grass AC Preamplifier Model #P511 and 9 cm electrodes were used to collect analog muscle expenditure data for the trapezius, posterior deltoid, bicep, and forearm flexor muscle groups. A Sound Pressure Meter was used to ensure a consistent dynamic of 76 + 5 db on all performances. An analog to digital converter, was used to convert analog data into digital form. The micro data were then integrated and rectified on an IBM 436 computer with a mathematics co-processor. Performances were recorded with two Audio Technica PRO 37R uni-directional cardioid condenser microphones on a Teak Stereo Cassette Tape Recorder Model V 270 and CrO2 recording tape.

    Students' performances were assessed on five dependent measures: (1)Rhythmic Accuracy, (2) Articulation, (3) Tone Quality, (4) Tempo Consistency, and (5) Intonation.  Scores for the criterion measures were determined by means of 7-point semantic differential scales that ranged from poor to excellent on each of the dependent measures.

    Spiccato and sautillé bowings were chosen for this study because they represent a refinement of basic "on the string," and "off the string" bowing strokes. The spiccato is an "off the string" stroke which utilizes a controlled bounce and is executed at the balance point of the bow. The sautille is an "on the string stroke" which utilizes an active bicep muscle and is executed more toward the middle of the bow. The sautille is used in fast passages when the controlled spiccato bounce is no longer possible. Subjects performed each series of tasks on a total of four sessions. Scalar patterns and melodic etudes were used as performance materials.

     Eight excerpts were performed using spiccato; and eight excerpts were performed sautillé. The spiccato and sautillé tasks were: three-octave scales, with four eighth-notes on each pitch, three eighth-notes on each pitch, one triplet on each pitch, three sets of string crossing in eighth notes, and two etudes. Etude I contained passages of four repeated notes on the same pitch; while Etude II contained single-note scalar and chordal passages. Spiccato excerpts 1 through 6 were performed at M.M quarter note=88; and, excerpts 7 and 8 were performed at M.M quarter note=92. Sautillé excerpts 1 through 3 were performed at M.M quarter note=144; and, excerpts 4 through 8 were performed at M.M quarter note=168. The sequence of tasks across sessions was identical for all subjects. Each session was at least 75 minutes in duration.  Subjects performed each of the 16 aforementioned excerpts three times.  Performances of the third trial on each excerpt at each session were audio-tape recorded and EMG potentials of the third trial were collected for subsequent analysis. Each excerpt was 30 seconds in duration, followed by a minimum rest period of 30 seconds. During each session, at least 25 minutes of additional time was devoted to instruction.

    Subjects were scheduled in the following manner: Sessions 1 and 2 were separated by a half-hour rest period. Sessions 3 and 4 were scheduled two days later, and were separated by a half-hour rest period. The close proximity of sessions 1 and 2 , and sessions 3 and 4 was arranged to save set-up time and to ensure more consistent electrode placement. To ensure that judges were blind as to which subject and session number (e.g., pre-post) they were hearing, pretest and posttest performance tapes were randomly ordered for evaluation. A panel of three judges, each having at least five years of public school or university string teaching experience, evaluated all performances at the conclusion of data collection. Separate ratings for pretest and posttest sessions were scored. Summated ratings for spiccato and sautillé bowing-types were calculated. Scores for rhythmic accuracy, articulation, tempo consistency, tone quality, intonation, and composite performance across eight spiccato excerpts and eight sautillé excerpts were summed to create composite spiccato and sautillé variables, respectively.

Procedure

      Subjects were randomly assigned to either an experimental or a control group. In order to ensure adequate learning curves, the length and sequence of experimental treatment were based on that used in other studies. All subjects received an orientation session during which subjects were asked to play a passage and to respond to an audio signal which monitored muscle activity in the targeted muscle groups, the trapezius and the forearm flexor. Subjects were asked to attempt to reduce the number of beeps by adjusting their muscle expenditure while performing. Because muscle expenditure is greater at high intensity than at low intensity, consistency in intensity was monitored throughout data collection by means of a sound pressure meter. Subjects were informed during performances when they were above or below the specified dynamic level. Therefore, an intensity range equivalent to 76 + 5 db was maintained for all bowing tasks in order to assure consistent dynamic level across bowings, sessions, and subjects.

      The experimental group (n = 24) participated in two 75-minute Biofeedback sessions and pretest (session 1) and posttest (session 4) assessments of muscle expenditure levels. The pretest and posttest assessment sessions were approximately one hour in duration. At each of the four sessions, subjects were asked to perform the 16 excerpts representing spiccato and sautillé bowing tasks described above. The same bowing tasks were used for each instrument, and for both treatment groups. EMG potentials on three consecutive trials for each bowing example were recorded during each of the four sessions. To determine baseline and terminal values, subjects received no feedback on the first and last sessions. On sessions 2 and 3, however, subjects in the experimental group received contiguous feedback in the form of an audio signal provided by a threshold beeper during each 30-second trial, and visual graph notation on the computer screen after each trial was completed. Between each trial, audio signal thresholds were lowered according to the response capabilities of the individual.  The researcher estimated the EMG threshold gain so that the auditory signal would sound a maximum of 25% of the trial time. Experimental group participants were instructed to reduce the amount of biofeedback beeps on each trial. Between excerpts, graphs using a Cartesian coordinate system for each of the four muscle groups were shown on the computer screen and discussed. For example, on sautillé excerpts, the bow arm often appeared fixed. A visual display of the trapezius muscle expenditure was shown in an effort to provide additional stimuli to increase the subject's knowledge of results.

    During the treatment sessions, experimental group subjects were asked to reduce tension levels on a specific muscle group. On the spiccato bowing (an off-the-string stroke), subjects were instructed to reduce forearm flexor EMG; while, on the sautillé bowing (an on- the-string stroke), the trapezius muscle group was targeted for EMG reduction. In addition to the biofeedback component of instruction, subjects were given verbal instruction relating to physiological aspects from the researcher to improve performance. Physiological information took the form of instruction in the movements required for successful execution of the bowings. For example, on sautillé strokes, it was mentioned that the bicep muscle must contribute more to cause the pumping motion of the elbow, thus contributing to the elasticity of the bow and causing the stick to bounce.

    The control group (n =23) participated in four private lessons each lasting 75-minutes.  At each treatment session, control group subjects, like the experimental group subjects, were instructed to perform the 16 excerpts which represent spiccato and sautillé bowing types. A rest period of approximately 30 seconds separated performance of each excerpt. The control group received the same amount of exposure to the bowing tasks as the experimental group. A traditional teaching approach which included physiological descriptions of the motions involved to aid in the improvement of each bowing task was employed. Pretest and posttest assessments of muscle tension levels in control group subjects were collected.  However, the control group received no EMG feedback during the treatment.

Biofeedback Data Analysis Procedures

     The analog data from the electrodes were processed by the myograph. The data were then converted to digital form using an analog to digital board. A total of 24,000 data points on each 30 second trial were recorded in micro form on the disk. The micro-data were then integrated and rectified using a Simpson Integration Program to determine millivolt-seconds over trials for each muscle group.

Results

     Inter-judge reliability among the three judges was determined for pretest and posttest on the five dependent measures of Rhythmic Accuracy, Articulation, Tone Quality, Tempo Consistency, and Intonation using the Cronbach's alpha . Reliability coefficients were high for all tasks and ranged between .94 and .98.

     Pearson Product-Moment Correlations were computed among the pretest and posttest muscle expenditure variables of forearm flexor, bicep, posterior deltoid, trapezius, and composite. Correlations were determined for the spiccato and sautillé bowings and were also broken down by experimental and control groups.  Correlations for experimental and control group muscle expenditure variables for the spiccato condition are shown in Table 1. Pretest correlations for experimental group muscle expenditure variables for the spiccato condition indicated that all muscles were significantly related with the exception of the correlation between trapezius and bicep muscles. Relations between composite muscle expenditure and trapezius, composite muscle expenditure and posterior deltoid, and composite muscle expenditure and forearm flexor were moderately strong. The correlation between trapezius and bicep was non-significant. On the posttest for the experimental group, several muscle expenditure variables were significantly related. The highest correlation was found between composite muscle expenditure and trapezius; while, the lowest correlation was found between bicep and posterior deltoid. Correlations between trapezius and forearm flexor, trapezius and bicep, and trapezius and posterior deltoid were non-significant.

     Pretest results indicated that subjects' use of a high degree of forearm flexor correlated highly with their overall muscle expenditure. Use of a high degree of forearm flexor may be the result of a fixation of the other muscles of the bow arm. On the posttest spiccato bowing condition, a strong correlation between the trapezius and composite, and lower correlations among the other muscle groups would be expected. This trend is reflected in the experimental group as these subjects progressed from pretest to posttest. The goal of the treatment was to lower forearm flexor muscle expenditure over trials. For the experimental group, the forearm flexor was highly correlated with the composite on the pretest. On the posttest, however, this correlation was much lower. Thus, following treatment, the degree of expenditure of forearm flexor tended to be independent of that of the other muscle groups.

      Pretest correlations for control group muscle expenditure variables for the spiccato condition indicated that expenditure measures for the trapezius, bicep, and forearm flexor were significantly related with the composite expenditure. In addition, forearm flexor and posterior deltoid expenditures were significantly related. While these correlations were statistically significant, the strength of relationships were low. The strongest relationship was found between composite muscle expenditure and trapezius. In contrast, the correlations between trapezius and bicep, trapezius and posterior deltoid, posterior deltoid and bicep, and forearm flexor and bicep were non-significant.

     On the posttest for the control group, all muscle expenditure variables were significantly related with the exception of those for trapezius and posterior deltoid, and trapezius and forearm flexor. While these correlations were statistically significant, the strength of the relationships was low. Moderate correlations were found between composite muscle expenditure and trapezius, and composite muscle expenditure and posterior deltoid. The negative correlation between bicep and trapezius, although low, is noteworthy. This indicates that as the bicep muscle expenditure increases, trapezius expenditure tends to decrease. In addition, there is an association between an increase in bicep expenditure and an increase in posterior deltoid expenditure. This explains the moderate correlations between composite muscle expenditure and trapezius, and composite muscle expenditure and posterior deltoid. On the pretest and posttest, the relations between the forearm flexor and composite were moderately low. The forearm flexor tended to be rather independent of the other muscles, but not to the extent that this was found for the experimental group.

     Pearson product-moment correlations for experimental and control group muscle expenditure variables for the sautillé condition are shown in Table 2. Pretest correlations for experimental group muscle expenditure on sautillé indicated that all muscle expenditure variables were statistically significant with the exception of the correlation between bicep and posterior deltoid, and bicep and forearm flexor.  Relations between composite muscle expenditure and trapezius, composite muscle expenditure and posterior deltoid, and composite muscle expenditure and forearm flexor were moderately strong. On the posttest for the experimental group, several muscle expenditure variables were significantly related. Moderately low correlations were found between posterior deltoid and bicep, posterior deltoid and forearm flexor, and bicep and composite. However, two moderately strong correlations were found between trapezius and composite, and posterior deltoid and composite, respectively.  Correlations between trapezius and forearm flexor, trapezius and bicep, trapezius and posterior deltoid, and forearm flexor and bicep were non-significant.  On the sautillé bowing condition, correlations among muscle expenditure variables were expected to be distributed predominantly between the trapezius and composite, and posterior deltoid and composite. This trend is reflected in the experimental group data.  In addition, results suggest that a high degree of forearm flexor use seems to be related to a truncated arm motion as a result of a lack of relation between bicep and posterior deltoid muscle expenditure. On the pretest, the relation between forearm flexor and bicep was not statistically significant. In contrast, on the posttest, this correlation was statistically significant and was moderately strong.  Pretest correlations for control group muscle expenditure for the sautillé condition indicated that relations between forearm flexor and posterior deltoid, and trapezius and composite were moderately high. Relations between composite muscle expenditure and posterior deltoid, composite muscle expenditure and bicep, and composite muscle expenditure and forearm flexor, although statistically significant, were low. Correlations between trapezius and forearm flexor, trapezius and bicep, trapezius and posterior deltoid, posterior deltoid and bicep, and bicep and forearm flexor were non-significant. On the posttest for the control group, moderately low correlations were found between trapezius and composite, posterior deltoid and composite. All other correlations were statistically significant with the exception of the correlation between trapezius and posterior deltoid.  On the sautillé bowing condition, the control group's muscle expenditure was fairly evenly distributed among the trapezius, posterior deltoid, and forearm flexor on the posttest.

     This trend is in contrast to the experimental group, whose muscle expenditure was distributed predominantly between the trapezius and composite, and posterior deltoid and composite. The relation between the forearm flexor and composite muscle expenditure was greater in the control group than in the experimental group.  Pearson Product-Moment Correlations were computed to examine relations among the performance measures of rhythmic accuracy, articulation, tone quality, tempo consistency, intonation, and composite performance for spiccato and sautillé bowings. These are broken down by pretest and posttest evaluation data, and by experimental and control groups.

    Pearson product-moment correlations for experimental group and control group performance measures on spiccato indicated that, on the pretest and posttest, strong correlations (r =.94 to.99) were found among articulation, tone quality, tempo consistency, intonation, and composite performance. In contrast, a significant negative correlation (r = -.50) was found between rhythmic accuracy and articulation in the experimental group on the pretest. This indicates that as articulation ratings increase, rhythmic accuracy tends to decrease. A significant correlation (r = .43) was found between rhythmic accuracy and tone quality in the experimental group on the posttest.  All other correlations with rhythmic accuracy were non-significant. Results indicate that rhythmic accuracy is a discrete measure of performance; while articulation, tone quality, tempo consistency and intonation appear to represent one other measure of performance.

      Pearson product-moment correlations for experimental group and control group performance measures on sautillé indicated that, on the pretest and posttest, strong correlations (r =.92 to.99) were found among articulation, tone quality, tempo consistency, intonation, and composite performance. In contrast, a moderate negative correlation (r= -.45 ) was found between rhythmic accuracy and articulation in the experimental group on the pretest. This indicates that as articulation ratings increase, rhythmic accuracy tends to decrease. In addition, a moderate correlation ( r = .50) was found between rhythmic accuracy and intonation in the control group on the pretest. All other correlations with rhythmic accuracy were non-significant. Results indicate, that for sautillé, rhythmic accuracy is a discrete measure of performance; while articulation, tone quality, tempo consistency and intonation appear to represent several performance measures which are themselves correlated.

    Relations among the performance measures of rhythmic accuracy, articulation, tone quality, tempo consistency, intonation, and composite performance for spiccato and sautillé bowings were consistent. For both pretest and posttest results, strong correlations (i.e., r =.92 to .99) were found among all performance criteria with the exception of rhythmic accuracy. The strength of the relationships did not vary according to experimental condition or task. Rhythmic accuracy had consistently low correlations with all other aspects of performance, indicating that subjects' performance of rhythms did not reflect the quality of performance in other areas. One possible reason for this finding is that the exercises are undulating in nature, and are not rhythmic. High interjudge reliability data supports this view.

Analysis of Variance

     To examine subject performances on composite performance as a function of the within subjects factor of pre-to-post test differences in performance, and the between subjects factors of group and instrument, data were analyzed by means of repeated measures analysis of variance procedures. The analysis of variance for Composite Performance appears in Table 3. The between subjects factor of group, the within subjects factor of pre-post, and the three-way interaction of group by instrument by pre-post were statistically significant. An examination of group means in Table 4 reveals that while significant pretest to posttest improvement in Composite Performance occurred regardless of group membership, the experimental group scored significantly higher than the control group on Composite Performance.

      The between subjects factor of instrument, interactions of group by instrument, group by performance, and instrument by pre/post, were all non-significant. However, the degree of improvement from pre to post varied significantly according to instrumental groups within the experimental and control groups as seen in the significant three-way interaction.  As can be seen in Figure 1, the high strings in the control group scored the lowest; while the low strings in the experimental group scored the highest on Composite Performance. In addition, high strings in the control group, and low strings in the experimental group made the most improvement from pretest to posttest. As is evidenced in Figure 1, the high strings in the experimental group and the low strings in the control group seemed to have improved the least. In the experimental group, the high and low string groups were relatively comparable in Composite Performance on the pretest with greater disparity on the posttest. The reverse was the case for the control group. To examine the magnitude of difference between experimental and control treatment conditions on Composite Performance, an effect size of .92 was computed.

Discussion

      In the experimental group, intercorrelations among muscle groups on spiccato from pretest to posttest changed considerably. A very strong correlation between composite muscle and trapezius on the posttest evaluation was found. This finding supports the view that the trapezius muscle should logically account for the most muscle expenditure on a spiccato stroke.  The trapezius must bear the weight of the whole arm, and the other muscle groups should function independently. The posterior deltoid should also act independently from the trapezius.  It follows that the strength of correlations between forearm flexor and trapezius changed dramatically from pre to post. The pretest correlation between forearm flexor and trapezius was significant, while no correlation between forearm flexor and trapezius was found on the posttest. This finding documents the independence of these two muscle groups on the posttest performance of spiccato.

     In treatment sessions of both groups, the goal of instruction was to improve performance and reduce tension in the forearm flexor by reducing the amount of wrist movement in the execution of spiccato strokes. The observed change in the strength of the correlations between the forearm flexor and the trapezius reflects this treatment goal. A number of subjects came to the first session with a spiccato stroke which appeared to incorporate a truncated arm and predominant wrist movement . This observation is supported by the higher pretest correlations in trapezius, posterior deltoid and forearm flexor with composite. Following treatment, there was a marked reduction in posterior deltoid, bicep and forearm flexor correlations with composite muscle expenditure on spiccato by the experimental group.

      A much higher correlation of posterior deltoid muscle expenditure was found in the control group in comparison to the experimental group. In the control group, correlations between forearm flexor and composite muscle increased slightly from a low correlation in the pretest to a moderate correlation in the posttest.  The lack of biofeedback in the control group could have accounted for the rise in forearm flexor expenditure from pre to post since this muscle was targeted for muscle tension reduction in the experimental group. The correlations between forearm flexor and trapezius were low on both the pretest and posttest. As evidenced in the experimental group as well, this finding shows the independence of these two muscle groups on the posttest performance of spiccato.

      There is a marked contrast between experimental and control groups in muscle group intercorrelations on spiccato muscle expenditure. The trend of muscle correlations by the control group appears to be the opposite to that of the experimental group. As the control group improved in performance achievement from pretest to posttest, muscle expenditure correlations decreased from a very strong correlation between composite muscle and trapezius on the pretest, to the moderately strong relationships between composite muscle expenditure and trapezius, and composite muscle and posterior deltoid on the posttest. While the forearm flexor (the targeted muscle group for mv/sec reduction on spiccato) was not strongly correlated with composite muscle expenditure, these muscle expenditure correlations did not decrease from pretest to posttest. A low correlation between forearm flexor and composite muscle is deemed desirable since minimal wrist movement is necessary for rhythmic accuracy, tempo consistency, and tone quality on spiccato strokes.

    In general, on sautillé, the expenditure is be expected to be more equally distributed among the trapezius, posterior deltoid, and bicep than on the spiccato; while on spiccato, the trapezius would bear the most expenditure. This view is supported by the data. On pretest correlations for the experimental group, muscle expenditure on sautillé indicated moderate relationships between the composite and a) trapezius, b) posterior deltoid, and c) forearm flexor. On posttest muscle expenditure of the experimental group, moderately strong correlations were found between the composite and a) trapezius, and b) posterior deltoid. The correlation between composite muscle expenditure and forearm flexor, however, decreased markedly on the posttest assessment of sautillé performance. Bicep correlations with composite also decreased slightly from pre to post. This is deemed desirable since the bicep acts oppositely from the forearm flexor. That is, as the forearm makes a movement to the right in executing the down bow, the bicep expenditure should decrease because the elbow moves slightly to the left to prepare for the rebound (up bow). This "pumping" motion of the upper arm increased from pre to post.

      For sautillé performance by the experimental group, correlations between forearm flexor and posterior deltoid were low on both pre and post. This is to be expected since the forearm and posterior deltoid should act independently. If they were not independent, a truncated arm motion which consequently raises trapezius muscle expenditure would result.

      In pretest control group correlations for sautillé, a strong relationship was found between composite muscle expenditure and trapezius. The muscle expenditure correlations between composite and posterior deltoid, composite and bicep, and composite forearm flexor were low, which indicated that total expenditure was not characteristically distributed among the trapezius, posterior deltoid, and bicep muscle groups. More distribution of expenditure is necessary for successful execution of the sautillé stroke. Also, the moderately high correlation was found between forearm flexor and posterior deltoid. This indicates a truncated bow arm which contributes to an increase in trapezius expenditure. Further, the lack of relationship between the bicep and forearm flexor on the pretest indicates that the bicep is not contributing the "shaking motion" necessary to execute the sautillé. This further places the burden on the trapezius.

      On posttest muscle expenditure for control group subjects in the sautillé condition, moderately strong correlations were found on composite muscle expenditure and forearm flexor, composite muscle expenditure and posterior deltoid, respectively.  These posttest correlations are deemed considerable improvements in that the muscle group expenditure is more distributed than on pretest evaluations. The correlations between forearm flexor and posterior deltoid decreased considerably from pre to post tests. This finding indicates increased independence among muscle groups, and a departure from the truncated muscle actions noted on the pretest. The correlation between forearm flexor and composite must still be considered to be higher than would be desirable. It would be hypothesized that the short down bow/up bow motions of the sautillé stroke by the control group are too static as a result of the degree of fixation among muscle groups.  This likely results in too much muscle expenditure by the forearm flexor in comparison to total expenditure. This in turn may be due to the lack of elasticity in the bow stick as a result of continually keeping the bow into the string rather than allowing the bow stick to jump.

      The repeated measures ANOVA analyses indicated that, while, significant pretest to posttest improvements in composite performance occurred regardless of group membership, the experimental group scored significantly higher than the control group on composite performance. It appears that biofeedback treatment had a more beneficial effect on performance for low strings. One explanation for this may be that the biofeedback treatment focused on improving kinesthetic awareness related to expenditure of the specific muscle groups which are responsible for optimal performance of spiccato and sautille bowing strokes. The execution of spiccato and sautille bowing strokes require undulating and energy efficient motions in order to achieve a high level of performance. It follows that kinesthetic improvements made in the execution of spiccato and sautille strokes as a result of the biofeedback training, in turn, brought about a significantly greater degree of improvement in composite performance.

      In summary, a primary goal of string teachers is to provide the most efficient instruction which in turn develops a sound and dependable technique in their students.  The student relies on the teacher for verbal instruction and modeling until the correct movements become spontaneous. String teachers often have difficulty in monitoring their students' physiological activity since observations of performance do not always provide sufficient information concerning the tension and release of muscles during performance. To examine physiological activity more closely, the string teacher often manipulates students' bow arms to help them feel the correct muscle expenditure and choice of muscle groups required for the execution of specific motions.

      Electromyography can be of great help to both students and teachers because it provides a means of quantifying the complex interactions among muscle groups involved in performance as well as improving kinesthetic awareness. Provision of string players with objective information related to muscle expenditure could help promote efficiency of movement and help modify undesirable physical behaviors in performance. The development of kinesthetic awareness is central to learning the vast array of motor skills involved in performance on a stringed instrument. Physiological aspects of string performance can be quantified and improved through the use of electromyography. Therefore, electromyography provides an additional means to develop kinesthetic awareness.

References

Cutietta, R. (1986). Biofeedback training in music: from experimental to clinical applications. Council for Research in Music Education , 87, 35-43.

Dennis, S. (1984). The effect of three methods of supporting the double bass on muscle tension. Journal of Research in Music Education , 32,95-103.

Dounis, D. C. (1921). The Artist's Technique of Violin Playing . Carl Fischer, Inc., New York.

Hodgson, P. (1934). Motion Study and Violin Bowing . J.H. Lavender & Co., London.

Irvine, I. K., Le Vine, W. R. (1981) The use of biofeedback to reduce left-hand  tension for string players. American String Teacher , Vol. XXXI, No. 3.

Koehler, W. K. (1989). The analysis of bowing skills through electromyographic biofeedback . Unpublished Research Paper.

Levee, J. R., Cohen, M. J., & Rickles, W. H. (1976). Electromyographic biofeedback for the relief of tension in the throat and facial muscles of a woodwind musician. Biofeedback and Self Regulation , 1, 113-120.

Le Vine, W.R. (1983). Behavioral and biofeedback therapy for a functionally impaired musician: a case report. Biofeedback and Self-Regulation , Mar. 8(1), 101-107.

Morasky, R. L., Reynolds, C., & Clarke, G. (1981). Using biofeedback to reduce left arm extensor EMG of string players during musical performance. Biofeedback and Self-Regulation , Dec. 6(4), 560-572.

Morasky, R. L., Reynolds, C., & Sowell, L. E. (1983). Generalizations of lowered EMG levels during musical performance following biofeedback training. Biofeedback and Self-Regulation , Jun. 8(2), 207-216.

Rosenberg, F. (1961). The Violin: The Technic of Relaxation and Power . Powell Publishing Co., Cleveland Ohio.

White, E. R., & Basmajian, J. V. (1974). Electromyographic analysis of embouchure muscle function in trumpet playing. Journal for Research in Music Education , 22, 292-305.

Zimmerman, F. (1966). A Contemporary Concept of Bowing Technique for the Double Bass . Leeds Music Co., New York.

Definition of Terms

A description of the muscle groups of the right arm which were utilized in this study

are as follows:

The Posterior Deltoid is the triangular muscle covering the rear shoulder joint which is used to raise the arm from the side.

The Trapezius is the large muscle of the upper back which controls the movement of the shoulder.

The Bicep is the large muscle at the front of the upper arm that flexes the elbow joint.

The Forearm Flexor is the muscle located at the inside of the forearm which is responsible for flexing the wrist and fingers.

The following bowing strokes were examined in the study:

The Spiccato bowing stroke is an off the string bowing which is executed by alternating down and up bows at the balance point of the bow for dynamics. When a "p" dynamic is called for, the bounce is executed closer to the middle of the bow.  Conversely, a dynamic is executed near the frog. This bowing utilizes a controlled bounce because the bow can be voluntarily rebounded.

The Sautillé bowing stroke is a rapid on the string bowing which depends on the bow stick for elasticity. This elasticity accounts for the off-the-string sound. The sautillé bowing utilizes an uncontrolled bounce which involves an involuntary rebounding action since it is performed at a tempo which is too fast for a deliberate controlled bounce.

This article is reprinted from the dissertation of the same name. Any further interests can be

directed to me via E-Mail:  wkkoehl@oratmail.cfa.ilstu.edu

These materials are copyrighted.

The Effect of Electromyographic Feedback

on Achievement in Bowing Technique

Dr. William K. Koehler

  Illinois State University

Abstract

This study examined the effect of electromyographic biofeedback training on achievement in spiccato and sautille bowing technique in a pretest/posttest design.  Forty-seven subjects (24 experimental, 23 control) participated in four 75-minute sessions. The experimental group received aural feedback, visual feedback, and traditional instruction; the control group received traditional instruction. Three judges assessed bowing achievement. Descriptive data on muscle tension, and performance achievement were collected.  Results indicated that both groups improved significantly from pretest to posttest on all performance measures. The experimental group and the low string subgroup in the experimental group scored significantly higher on all performance measures. Following treatment, forearm flexor expenditure tended to be independent of the other muscle groups on spiccato. For sautille, significant correlations between the forearm flexor and bicep were found in both groups on the posttest.