Abstract
The pedagogical literature strongly suggests that when a musician works on a specific piece of repertoire, one should choose practice strategies tailored to the challenges presented by that material. Such a behavioral choice could represent an instantiation of focused and deliberate practice, a critical aspect of the relationship between experience and the acquisition of expertise. However, most of the literature investigating individual practice has used a single stimulus for all participants, or else has employed surveys or other proxy measures of practicing behavior. In this study, participants of three different experience levels (high school, collegiate, or professional violinists) practiced three excerpts, each featuring a different signature challenge. Results suggest that practice is highly idiosyncratic, that participants do adjust their approaches to the challenges of the material, but that individuals of differing experience levels identify remarkably similar problems within the material.
Private, individual practice is arguably the central experience of musicians’ lives. Some individuals begin their musical lives through private lessons, others through school music programs; some are self-taught. The uniting feature of all these paths is the time musicians spend practicing and refining the skills learned in these various settings through individual work.
Accumulating vast amounts of deliberate practice is a central component in the acquisition of expertise in many domains, including music (Ericsson, 2008; Ericsson et al., 1993; Ericsson & Pool, 2017; Lehmann & Ericsson, 1997). The idea that 10,000 hr of focused work are a necessity in developing a skill area to an expert level has even entered the popular science press, although researchers stress that the importance of the particular number has been exaggerated in popular works (Ericsson, 2012; Ericsson & Pool, 2017; Gladwell, 2011). However, critics have noted that aggregate practice time is insufficient to explain differences between individuals’ ability levels (Hambrick & Meinz, 2011b). In a variety of domains, both individuals’ interests and stable, heritable personality and cognitive traits play a great role in determining their success (Hambrick et al., 2008, 2010; Meinz et al., 2012). Even among professional musicians, such stable traits (that could collectively be called “talent”) play a role in determining individual performance levels (Campitelli & Gobet, 2011; Hambrick & Meinz, 2011a; Meinz & Hambrick, 2010). Across time ranges spanning single practice sessions to a college semester to a career, the amount of time that musicians spend practicing appears not to be the only, or sometimes even the greatest, predictor of their achievement (Duke et al., 2009; Jørgensen, 2002; Madsen, 2004).
The apparent disagreement between research indicating that large quantities of lifetime practice are necessary to become an expert musician and those studies that find a limited connection between practice and achievement may not present a paradox, however. Although factors such as working memory capacity may play a role in determining whom among those individuals who have chosen to engage in a given field will ultimately reach the highest levels in the most efficient manner, accumulating a vast amount of sheer practice time is a prerequisite to expertise (Gladwell, 2011; Meinz & Hambrick, 2010). Gladwell (2013) described practice’s prerequisite role elegantly in The New Yorker: “In cognitively demanding fields, there are no naturals. Nobody walks into an operating room, straight out of a surgical rotation, and does world-class neurosurgery.”
A second factor mediating the relationship between accumulated practice experience and expertise is the nature of the practice in question. Ericsson (2012) stresses that it is deliberate, goal-directed practice that develops skills and expertise. The nature of what deliberate effort looks like in musical practice has been a frequent topic in the pedagogical literature even before researchers began explicitly probing the phenomenon. Horn pedagogue Phillip Farkas (1956) wrote, “In every difficult passage, there is one element or more which accounts for its difficulty . . . Whatever the difficulty, it is the player’s duty to discover and conquer it” (p. 45). Galamian (1985) clarified the process of discovery and conquest, specifying that the player must determine the nature of each technical challenge (e.g., shifting, bowing patterns), then “[isolate and reduce it] to its simplest terms so that it will be easier to devise and to apply a practice procedure for it” (p. 99). Many books identify connections between various musical and technical challenges and specific practice activities intended to remediate each one, with some even taking the form of a diagnostic manual (Fischer, 2004; Gerle, 1983, 1991; Kish, 2017; Nardolillo, 2015; Wye, 2000). The pedagogical literature thus provides one clear model of the nature deliberate of musical practice, a model that centers on analyzing problems, utilizing strategies specific to each challenge, and ultimately eliminating them as obstacles to the desired performance.
Practice has been a topic of growing attention within the music education research literature. Duke et al. (2009) found no correlation between total practice time and performance level on a retention test the next day; instead, the highest performing individuals exhibited a common suite of behaviors found in its entirety among no lower scoring participants. Professional and graduate students appear more prone to target particular problematic sections than undergraduates, and experience level also correlates positively with the use of sophisticated practice strategies that reflect individuals’ awareness of their own strengths and weaknesses (Hallam, 2001a, 2001b; Maynard, 2000, 2006). Indeed, self-regulation—the ability to monitor one’s own actions and progress and to make adjustments accordingly—seems to develop alongside musical experience (Austin & Berg, 2006; Christensen, 2010; McPherson & Renwick, 2001; Miksza, 2006, 2011, 2012). While beginning musicians seem to know few practice strategies and to use them infrequently, their use among experienced musicians has been found in some cases to correlate more strongly with age and experience than with performance scores (Christensen, 2010; Hallam, 2001a; Pitts & Davidson, 2000). Experienced, high-performing musicians also incorporate musical interpretive details earlier in practice than others (Chaffin & Imreh, 2001; Duke et al., 2009; Hallam, 2001a).
Rests early and late in the practice session seem to facilitate non-pianist musicians’ learning of short sequences on the piano (Cash, 2009). Early rests also seem to facilitate retention at performances the next day, as does distributing practice across multiple days and across sleep intervals (Cash, 2009; Simmons, 2012; Simmons & Duke, 2006). Listening to a model recording can facilitate rhythmic accuracy and performance speed gains over a practice session (Cash et al., 2014; Henley, 2001; Rosenthal, 1984). As in other performance areas such as athletics, focusing the abstract, distal consequences of action (e.g., focusing on the sound) leads to more concrete gains in practice than focusing on proximal actions (e.g., focusing on the piano keys; Duke et al., 2011).
The research literature in music practice seems to support the pedagogical model of deliberate practice as a goal-directed process consisting of sophisticated choices driven by analysis of the intended outcomes of one’s practice, in conjunction with an awareness of one’s own strengths. However, most of the studies cited above examined musicians’ practice only of a single musical task or excerpt. Some employ interviews and self-reporting rather than observations of actual practice, define sophisticated practice strategies a priori rather than through identifying what works in an actual practice session, or employ case study procedures. These limitations preclude the assessment of whether musicians’ behaviors actually support pedagogues’ assertion that musicians analyze the different types of challenges in the material they are learning and pick different behavioral strategies and approaches accordingly.
In this study, I sought to determine whether musicians do indeed display different practice behaviors as a function of the type of material they are learning. I sought answers to two specific questions: Do violinists employ different observable behaviors when practicing material containing categorically different challenges, and if so, do their approaches to each type of challenge vary as a function of their experience level?
Method
Participants
Violinists at the high school (n = 11, female = 8), collegiate (n = 12, female = 4, undergraduate = 10), and professional (n = 12, female = 3) levels were recruited to learn three excerpts in consecutive 10-min practice sessions. Of the 12 high school participants originally recruited, one did not complete the study procedure until after entering college and ceasing regular practice several months later, and was therefore excluded from all analysis. Collegiate participants included nine violin performance majors, two music education majors, and one Bachelor of Arts in Music major who had been accepted to a master’s program in violin performance. Professional participants were selected by their acquisition of and continued employment in professional-level positions through competitive processes, and included full-time university performance faculty and chamber musicians, as well as professional symphony members. Student participants were recruited both directly and through their teachers, and consideration was given to selecting participants who possessed the skills needed to learn the excerpts in the required time frame.
Stimuli
To inspect participants’ behavioral responses to different kinds of material, I composed three excerpts for the study, each saturated with a particular technical challenge endemic to violin playing. One excerpt required many shifts, another featured copious string crossings, and the third employed a syncopated bowing pattern across the entire excerpt. Based on pilot participants’ experiences, I presented high school participants with a version of the shifting excerpt transposed down a third but preserving all the relevant leaps. Because I was interested in examining musicians’ behaviors as they would be deployed in practice of learning regular musical material, the excerpts were designed as authentic melodies including typical figures and implied harmonies. As such, each excerpt includes possibilities to employ the signature technique of one of the other excerpts. For instance, some participants chose to start the slurring excerpt in second position, and while practicing the string crossing excerpt, others experimented with smaller shifts coupled with string crossings (Figure 1). However, each excerpt is clearly defined by copious use of its own particular challenge.
The (A) slurring, (B) string crossing, and (C.1) shifting excerpts, including the (C.2) high school variant, transposed down by a third but maintaining the need for each shift. Although no excerpt was devoid of either of the other challenges, and participants’ execution choices (e.g., starting C.1 in second position) could somewhat alter the mix of challenges the material presented, each excerpt was overwhelmingly saturated with one signature problem to be solved.
Procedures
After giving informed consent and receiving initial instructions, each participant participated in three consecutive 10-min practice blocks, one for each excerpt. Participants answered small sets of background questions, serving as a short rest between blocks; questions following the final block included inquiries into their experiences in the study. Initial instructions informed participants that the purpose of the study was to examine how musicians practice; that they would learn three different excerpts in 10-min practice sessions immediately followed by performance; and that their task in each 10-min session was to “get each excerpt as close to the target tempo as possible while playing well.” I provided a pencil and a metronome, although some elected to use their own equipment.
Each block involved several steps. Participants received a copy of the excerpt they were about to practice, which they could examine while listening to a computer-generated model recording three times. I reminded them that the goal was to play as close to the target tempo as possible while still playing well and that they were free to use the provided metronome and pencil as much or as little as desired. They then practiced the excerpt for 10 min. I remained in the room with them as they practiced and gave warnings when they had 5 min and 1 min remaining. Rather than interrupting participants’ work, I waited until they paused near or slightly after the 5-, 9-, and 10-min marks to issue warnings or end the practice session. Finally, the participants performed the excerpt three times, I collected their sheet music, and we proceeded to the group of questions following that block. The performances were intended simply to provide a goal during the practice session; participants’ actual performances were not assessed. I made a video recording of all of participants’ activities, including initial instructions, all three blocks, and subsequent questions for later analysis.
The model recordings were generated using Finale 2008 and the Garritan Personal Orchestra, and were intended to partially mitigate differences in sight-reading abilities while also setting an extremely fast target tempo for each excerpt. Three hearings were certainly not enough to allow any participant to memorize the material, but served to demonstrate the key, meter, and general character of each excerpt. The fast model recording tempi in combination with the reiterated instructions were designed as a self-regulated difficulty adjustment, facilitating the use of common excerpts across a wide range of experience levels. The extreme tempi set a high bar even for professional musicians; the instructions gave all participants explicit permission to slow down, with the primary goal being playing well. These instructions were given in the initial directions and repeated before each practice session.
Data Transcription and Analysis
Both research questions address observable changes in practice behavior in response to differences in the material participants were learning or their experiences. Although I anticipated seeing practice behaviors such as those observed in prior practice studies (e.g., Duke et al., 2009; Maynard, 2000, 2006; Miksza, 2007) or prescribed in the pedagogical literature, I wished to avoid limiting my observations to a list of activities defined a priori. Instead, I designed a procedure to identify and extract frequencies for behaviors as participants exhibited them in practice.
I watched each video and transcribed each behavior I saw using F4 transcription software, developing a system to efficiently describe what happened at each point in the video. Most of the footage consisted of participants playing part or all of an excerpt, stopping, and then playing again from the same or another point. I labeled these moments as “playing events” and labeled each event’s first and last note, and the software provided the stop time in the video, whereas the start time could be determined from the previous event. I also described any other potentially salient details such as pauses, changing tempi, wrong notes, altered rhythms, and other deviations from the notated material. A sample playing event from a slurring excerpt practice session appeared in the transcript as: “Play: 5.1.4 to 6.4.1. Alters rhythm: same as previous event. Plays 5.2.3 as an E rather than an F#. <00:25:04>”
I defined rules to demarcate boundaries between playing events using behaviors that appeared to represent a clear new unit of activity. When participants stopped progressing forward through the excerpt and returned to an earlier point in the score, even by a single note, I created a new playing event, unless it were part of an entire event consisting of playing material in retrograde. I developed labels to describe both systematic alterations to the printed material (e.g., altered rhythms) and apparent mistakes. Participants rarely skipped forward in the score, in which case I also recorded a new event, unless it was part of a systematic, observable omission pattern (e.g., omitting the low voice in the shifting excerpt). Nonplaying events included marking the part, adjusting the metronome, and other nonplaying activities, and were readily distinguished in the video record. This procedure generated nearly 12,900 discrete events described in text files.
Labeling which notes within the excerpt served as the first and last performed note of each playing event allowed me to compare how much material (as a percentage of the excerpt) participants included in each playing event, how frequently each note served as the initiation point of a playing event, and how frequently each note was played across an entire practice session. I also analyzed the overall length in seconds of each practice session; the percentage of time in each session spent playing as opposed to any other activity; the number of playing events in each session; the number of complete, beginning to end playing events in each session; and the amount of time in seconds during which the metronome was active.
To collect the systematic alterations that I identified in my transcripts, I wrote a Python script to consolidate and count descriptors’ appearances by abstracting specific details into more general tags (e.g., replacing references to specific notes with <<notetag>>). Events included multiple tags as needed to describe participants’ engaging in multiple deviations from a simple attempt at a section of the material. I ran a second script iteratively, one that collapsed textual variations that referred to the same behavior into a single code (e.g., “repeats <<notetag>> several times” and “repeats <<notetag>> many times”). After running the script many times, 14 codes emerged that occurred in at least 0.5% of events (65 events) in the data and therefore appeared to represent systematic practice patterns, but that were not consolidated into other codes: repeats (a) first, (b) last, or (c) an internal note; (d) marks part; (e) adjusts metronome; plays with systematically altered (f) rhythms, (g) bowings, or (h) added double stops; plays (i) pizzicato rather than arco, (j) in a different octave, (k) open strings only, or (l) backwards (i.e., retrograde through the music); (m) includes a note in the slur that precedes it; and (n) plays with the metronome active but at a different tempo.
While transcribing participants’ practice videos, I identified two more behaviors of interest that both involved patterns over multiple playing events rather than observed within a single event. First, repeated, clustered attempts at the same subset of the material were present in many practice sessions. Several prior researchers (e.g., Miksza, 2007) have identified whole-part-whole practice as a recognizable strategy, and Maynard (2000, 2006) identified practice frames, modeled on the rehearsal frames defined by Duke (1994), that likewise center around concentrated, sustained effort on a single practice target. I defined a detail frame as a series of three or more playing events that had in common a subset of the material that included no more than 20% of the excerpt and that could include marking the part or adjusting the metronome, but not playing other material.
I also noticed a pattern in which two or more consecutive playing events, if viewed as a whole, could be construed as a single attempt at an extended portion of the material or of the complete excerpt, interrupted by one or more small backtracks of just a few notes, often following an incorrect note. I labeled such sequences “ratcheted practice” (after the tool that registers motion in only one direction), and included an event in such a sequence if (a) it and the event before or after it were not separated by nonplaying events, (b) it and that other event overlapped in the score (c) by no more than one beat, (d) it was not part of a detail frame, and (e) it and the other event both individually satisfied the requirements for labeling as a ratcheted practice event. Also, (f) playing events in which a participant reached the end of an excerpt concluded a series of ratcheted events, even if the subsequent event would otherwise qualify (i.e., if the participant repeated the last beat), as reaching the end seemed to preclude the possibility that this repetition would be perceived as continuous playing. Series of three or more ratcheted events were sometimes present when an extended performance trial was interrupted more than once and each of the events individually satisfied the requirements.
Reliability
A tenured professor of violin assessed 20% of the practice sessions recorded in the initial data gathering. Rather than asking this professional to re-extract each performance event, I provided a Scribe file that segmented each section into events, as well as a scoresheet, on which the reliability assessor used brackets to indicate what material (if any) was played and checked boxes indicating which of the behaviors of interest were visible in each clip (Duke & Stammen, 2011). After consultation to resolve differences arising in part because of the different rounding methods for time used by Scribe and the software used for initial transcription, we agreed on 92.7% of events.
Results
The questions under investigation in this study were whether violinists modulate their behaviors in response to the nature of the material they are studying, and the extent to which any such modulations differ between participants of varying experience levels. The design incorporated two independent variables: excerpt focus (shifting, syncopated slurs, or string crossings) and experience level (high school, college, or professional). It also yielded many dependent variables, including the frequencies of the 14 practice behaviors extracted from the narrative descriptions, what and how much material was played in each playing event and across the whole session, the number of playing and other events in each practice session, the duration in seconds of the practice session and each event, the number of complete performance trials in each session, the time the metronome was active, and the data about which events were parts of series of events labeled as detail frames or ratcheted practice. To compare behaviors between individuals, the data for each behavior was measured as counts and frequencies across entire practice sessions. With two independent variables and many dependent variables, frequency data were used to assess the degree to which these observable behaviors varied depending on what material was being practiced and the participants experience level.
What Material to Practice
Perhaps the simplest way to assess whether violinists approach different material using different strategies is to look at what they focus on. Figure 2 shows (thin lines) the frequency with which participants in each group selected each note of the three excerpts as a starting point for a playing event and (thick lines) the cumulative amount of attention devoted to each note, as measured its percentage of all notes played in the practice session. All three groups’ curves in each part of Figure 2 are remarkably similar to one another, yet pointedly different between excerpts. That fact that graphs differ between excerpts may simply point to their composition, indicating where the difficult sections lie, but the similarity between groups suggests few differences as a function of experience.
Participants’ attention allocation within each excerpt, as measured by their choices of starting locations within the material (thin lines, left Y axis) and their aggregate practice of each note within the excerpt (thick lines, right Y axis), represented as each note’s percentage of all notes played within each practice session. Note that the scale in the slurring excerpt (A) is different than the other two graphs due to the frequency with which participants started at the beginning: (A) slurs, (B) string crossings, and (C) Shifts.
Participants in all three groups overwhelmingly chose to start at the beginning of the slurring excerpt, almost all playing events start at the beginnings of measures (the labeled notes on the X axis in Figure 2), and the overall amount of time they spent practicing any particular note was fairly evenly distributed across all of the material. When practicing the shifting excerpt, participants often started at the beginning of the excerpt or at the beginning of other measures, but they also frequently started at locations immediately before or after a shift, such as notes 7, 9, 30, and 78. Their accumulated practice shows that they spent much of their effort on the very beginning of the excerpt, the shifts at the end of Measure 2 and the beginning of Measure 3, and the final four measures, but they completed fewer repetitions of the notes in Measures 4, 5, and 6. When practicing the slurring excerpt, participants chose to start almost all of their playing events at measure boundaries rather than in the middle of measures, but they also focused their attention on specific locations, especially Measures 3 to 4 and the final six measures of the excerpt.
Overall, participants’ choice of what material to play suggests a focus on continuity and maintaining the pattern when they practiced the slurring excerpt, starting at the beginning and other musically sensible locations, and distributing their attention across the whole excerpt. When practicing the shifting excerpt, though, they started at and devoted much of their attention to specific difficult locations, even when they did not begin a measure. And when practicing the string crossing excerpt, they struck a balance between approaches, working again for continuity across passages, passages that were more narrowly defined than in the string crossing excerpt but less discrete than in the shifting one. The data do not, however, provide much evidence that participants of differing experience level made different decisions about what to work on.
Practice Behaviors
Data for how participants practiced each excerpt, as opposed to what within the material they worked on, likewise show differences as a function of the material. The average amount of material covered in each playing event was longest in the slurring excerpt practice (high school = 26.5%, SD = 10.3%; college = 33.8%, SD = 10.7%; professional = 40.2%, SD = 11.1%), intermediate when participants practice the string crossing excerpt (high school = 18.6%, SD = 6.30%; college = 24.6%, SD = 8.68%; professional = 29.0%, SD = 8.74%), and lowest when they practiced the shifting excerpt (high school = 16.5%, SD = 8.06%; college = 12.7%, SD = 3.08%; professional = 13.6%, SD = 4.61%). Most participants executed at least one complete, beginning to end performance trial when practicing the slurring excerpt (12 of 12 professional, 11 of 12 college, and eight of 11 high school) and the string crossing excerpt (12 of 12 professional, 10 of 12 college, and seven of 11 high school), but this was less common during shifting practice (five professionals, three college and high school). Among those participants who executed a complete performance trial at least once in each setting, those events represented a higher percentage of all playing events in the slurring context (high school = 8.08%, SD = 9.39%; college = 12.5%, SD = 9.02%; professional = 17.0%, SD = 11.3%) than in other settings. Among college and professional participants, practice of the string crossing excerpt also included a higher percentage complete performance trials (college = 7.63%, SD = 8.01%; professional = 9.80%, SD = 7.53%) than did practice of the shifting excerpt (college = 1.0%, SD = 0.37%; professional = 2.96%, SD = 2.90%). The fewer high school participants who exhibited them instead recorded a similar percentage of complete performance trials when practicing the string crossing (2.38%, SD = 1.72%) and shifting (5.05%, SD = 4.33%) excerpts.
All participants engaged in at least one detail frame in all practice sessions. The percentage of events that occurred within these frames was lowest when participants practiced the slurring excerpt (high school = 31.1%, SD = 13.3%; college = 30.2%, SD = 17.3%; professional = 19.2%, SD = 12.0%). Collegiate and professional participants also included a lower percentage of events in detailed frames when they practiced the string crossing excerpt (college = 54.9%, SD = 14.7%; professional = 52.5%, SD = 14.9%) than when they practiced the shifting excerpt (college = 65.7%, SD = 15.1%; professional = 68.7%, SD = 9.34%). Once again, high school participants’ rates were similar when they practiced string crossings (58.5%, SD = 22.6%) and shifts (58.1%, SD = 18.2%).
The number of playing events per session was the lowest among college and professional participants when they practiced the slurring excerpt (college = 83, SD = 26.5; professional = 73, SD = 21.7) but was comparable between the string crossing (college = 115, SD = 39.6; professional = 115, SD = 40.5) and shifting (college = 115, SD = 41.3; professional = 127, SD = 48.6) practice sessions. High school students completed similar amounts of playing events to the other two groups when they practiced the string crossing (113, SD = 39.3) and slurring (87, SD = 35.2) excerpts, but their number of playing events in shifting excerpt practice sessions (89, SD = 41.1) was similar to that of the slurring rather than the string crossing practice sessions.
Ratcheted practice occurred in every practice session among student participants except for a single college student’s string crossing practice. Among professionals, only half displayed ratcheted practice while learning the string crossing excerpt, but it was frequent in other settings (shifts: 12 of 12, slurs: 10 of 12). Among those individuals who displayed ratcheted practice behavior, it occurred in a smaller percentage of events when college and professional participants worked on string crossings (college = 6.25%, SD = 3.58%; professional = 3.94%, SD = 3.00%) than when they worked on shifts (college = 16.3%, SD = 11.9%; professional = 11.1%, SD = 6.53%) or slurs (college = 14.6%, SD = 10%; professional = 8.93%, SD = 5.81%). High school participants were different, displaying comparable rates of ratcheted practice in all practice contexts (string crossings = 16.4%, SD = 14.7%; shifts = 15.2%, SD = 6.59%; slurs = 19.1%, SD = 13.7%).
For all groups, the percentage of time spent playing was similar when they practiced string crossings (high school = 84.9%, SD = 8.71%; college = 86.5%, SD = 6.04%; professional = 85.9%, SD = 6.81%), shifts (high school = 82.8%, SD = 4.0%; college = 84.8%, SD = 5.04%; professional = 80.8%, SD = 9.30%), and slurs (high school = 88.7%, SD = 7.36%; college = 84.6%, SD = 6.67%; professional = 78.6%, SD = 12.3%). Because I did not cut off their work at the 10-min mark but instead allowed participants to finish their activity, and because a few participants chose to skip to the performance before their time had expired, I also compared the duration of each practice session. Participants’ time spent in seconds was similar between string crossing (high school = 630 s, SD = 14.2 s; college = 604 s, SD = 25.2 s; professional = 598, SD = 61.2 s), slurring (high school = 612 s, SD = 65.1 s; college = 606 s, SD = 38.2 s; professional = 626 s, SD = 92.8 s), and shifting (high school = 646 s, SD = 25.8 s; college = 648 s, SD = 41.1 s; professionals = 615 s, SD = 40.1 s) practice sessions.
Many of the other measured behaviors seem to necessarily involve some level of conscious decision-making, and thus could be considered “practice strategies.” Figure 3 illustrates common patterns among these behaviors. Individuals’ use of each of these behaviors was highly idiosyncratic, the clearest example being playing material pizzicato rather than arco. Despite its accumulating enough instances to be included for analysis, pizzicato playing characterized the practice of just four individuals (two professionals and one in each student group) and hence is not included in Figure 3. Substantial groups of participants have abstained completely from many other behaviors, too, while others displayed them frequently.
Frequency of various overt practice behaviors. Most overt behaviors are characterized by their presence or absence within one practice context, though some show degrees of scale; most also show considerable idiosyncrasy in whether they were or were not used at all.
The second overall trend seen in Figure 3 is that most participants either employed each practice strategy only in practicing one excerpt, or else they employed it on two excerpts while abstaining on the third. When practicing the shifting excerpt, more participants engaged in playing backwards and octave displacement, and they did so for a larger percentage of playing events, than in other practice sessions. They mostly abstained from altering the rhythms and the bowings, they used and adjusted the metronome less, and four professional participants ignored the metronome and played a different tempo during practice of the string crossing excerpt than others. They started at the beginning of the excerpt more often and adjusted their metronome more frequently when they were practicing the slurring excerpt than in other practice sessions. They also “included a note in the previous slur” in this context, though this was less clearly a conscious decision. When practicing the string crossing excerpt, participants were more likely to add double stops than elsewhere (though many also added double stops to the shifting excerpt), and they altered the rhythms and to some extent the bowings.
Participants marked their parts at comparable rates across excerpts, and participants with more experience appeared to do so somewhat more often except while practicing string crossings. Professionals adjusted the metronome more and included a note (non-systematically) in a preceding slur less than others, although these behaviors may reflect performance accuracy rather than conscious decisions. Some professionals also ignored the metronome in the shifting excerpt and played a different tempo, and more professionals than others played backwards in the shifting excerpt and isolated the open strings in the string crossing excerpt. High school students used the metronome less frequently than the other groups, whereas college students added double stops frequently.
Finally, I recorded three distinct repetition behaviors: repeating single notes at the beginning or end of a playing event, or repeating an internal note (without backtracking and triggering a new event). All participants displayed every type of repetition while practicing the shifting excerpt with the exception of a single professional who did not repeat any internal notes. One professional and one high school student refrained from repeating any first notes while practicing string crossings; 33% of college students and 25% of professionals abstained from repeating any last notes; and 18% of high school students, 25% of college students, and fully 50% of professionals abstained from repeating internal notes in this setting. About 36% of high school and 17% of college students, but no professionals, abstained from repeating first notes while practicing the slurring excerpt; 45% of high school and 33% of college and professional participants refrained from repeating last notes; and 27% of high school students but fully 50% of college and professional participants repeated internal notes.
Discussion
When presented with three excerpts that each featured a different technical challenge, participants in this study appeared to practice each one using a different behavioral suite, though they displayed each behavior idiosyncratically. As seen in Figure 3, practice sessions of each excerpt tend to be typified by the absence or presence of each measured practice strategy. One could reidentify an unlabeled practice session that featured frequent systematically altered rhythms and bowings, a high percentage of playing events starting from the beginning, few added double stops, and frequent metronome adjustments as representing a participant’s work on the slurring excerpt.
Behaviors that less clearly represented conscious decisions, such as the percentage of playing events in detail frames or ratcheted practice, the average length of a playing event, and the frequency with which participants executed complete, beginning to end performance trials during practice also provide evidence of differing goals in each practice session. Playing events covering longer average portions of the excerpt and including more complete performance trials, low percentages of work inside detail frames, and the presence of ratcheted practice suggest that participants worked for continuity across the material when they practiced the slurring excerpt. When they practiced the shifting excerpt, however, they played much smaller chunks of material on average, included a higher percentage of these repetitions in detail frames, and devoted more of their attention to specific, difficult locations within the excerpt at the expense of other measures, rather than spreading their attention evenly. When learning the string crossing excerpt, they displayed something of a hybrid approach. Their repetitions were longer than when they practiced the shifting excerpt, but they engaged in more detail framework than in the shifting excerpt. Their aggregate attention, as seen in Figure 2, was concentrated in certain spots, but these spots were broader than in shifting excerpt practice, often encompassing several measures. It appears they were working for continuity, but across smaller sections than in the slurring excerpts, and as in their shifting work, they appear to have identified these sections as noticeably more difficult and attention worthy than other parts of the excerpt.
The three experience level groups, however, were very similar in the specific points in the material they chose to focus on. Whether a spike in Figure 2’s aggregated practice charts occurred in one or another measure was inconsequential, reflecting the stimulus’s composition rather than meaningful participant decisions. However, the fact that nearly every location in the material where one group spent extra effort was closely paralleled by the other groups’ accumulated practice is remarkable. A well-established phenomenon in the study of expertise is that experts perceive and recall domain-specific stimuli differently, seeing a few structures while others perceive many discrete elements, as when a chess grandmaster is able to reconstruct a real-game board configurations using fewer glances at the model board than less expert players. In music practice, one would expect that experts might perceive the challenges in the material inherently differently than less experienced participants and therefore might work on slightly different material, but there is little evidence for such differential problem identification as a function of expertise in this study.
One possibility is that while professional violinists are more expert performers than students, they are not necessarily better at practice. Their years of cumulative practice and greater experience with similar, previously learned material perhaps allow them to perform at the professional level, even the efficiency of their practice relative to their experience level was no different than that of students. Previous research has suggested that teachers often attend to what to practice, the product, or the repertoire and the musical changes to be made, rather than how to go about making those changes. Specific high school participants were recruited who possessed the position skills to make reasonable progress on the shifting excerpt, and who may have had a higher experience or expertise level than the typical high school student, perhaps also contributing to the groups’ similarity.
Another possibility is that the excerpts were so short and so saturated in their respective technical challenge that even the least expert participants were easily able to identify what they were “supposed to” focus on. The excerpts were designed to maximize the distinction between different contexts, rather than to require participants to tease out subtle, embedded challenges. Indeed, the one hint in Figure 2 of experts’ possibly identifying a challenge differently can be seen in the first measure of the shifting excerpt. There, all three groups display a spike in both the cumulative attention and starting point data. However, among student participants, both of these spikes occur on Subdivision 9 (the note following the shift), whereas professional participants’ spikes occur on Subdivision 7 (the two sixteenth notes preceding the shift), and professionals’ aggregate attention spike continues across the shift. This would suggest that at this one location, professionals perhaps perceived the shift as the difficulty, whereas students instead identified the high-register material following it as the problem to work on. One instance is weak evidence, but it occurred at the first location in the material where a shift was required, and perhaps participants accumulated these different practice patterns before all groups had come to realize shifts were the excerpts’ major challenge. Further research employing less heavy-handed stimuli could further explore whether and how expert musicians differ from students in their perception of problems within the material.
Teachers and students may be interested in whether they should deploy a specific behavior in response to a specific challenge. For instance, should a student employ more altered rhythms when practicing string crossings? The present data do not address whether such behavior did in fact facilitate acquisition or retention of performance improvements, only whether participants decided to employ the behavior or not. However, outside of repeated notes, most practice behaviors do appear to typify certain practice sessions by their relative frequency; a preponderance of the evidence suggests that violinists choose how to practice material strategically, or at least behave as if they believe their behaviors are strategic. Future studies designed to distinguish actually efficient behaviors from merely employed ones (i.e., those that effect lasting performance changes in fewer repetitions or less time than other behaviors) may require more focused, statistically assessable methods.
This study also does little to distinguish any mechanism explaining why the various practice behaviors examined should be effective. When the author was a violin student, for instance, his teacher suggested that playing with altered rhythms allowed the brain to consider technical material from a variety of perspectives. An alternative explanation is that systematically altering the rhythms only enforces a minimum number of repetitions. Future studies could examine the relative efficiency of a variety of practice strategies with consideration of the number of repetitions of the material required to complete the behavior, thus determining whether these strategies represent a truly different approach or simply a regime to enforce repetition.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.