The influence of an extended music curriculum at school on academic self-concept in 9- to 11-year-old children

Abstract

Cognitive transfer effects of music lessons on several cognitive abilities such as IQ (Schellenberg, 2004) or language skills (Moreno et al., 2009) have been reported. Beyond that, also conative transfer effects (i.e., motivational aspects like perseverance) of music lessons have been revealed. One such conative transfer has been found for academic self-concept (Degé, Wehrum, Stark, & Schwarzer, 2014). Self-concept describes the evaluations a person holds about him/herself. However, this study was correlational. Hence, it remains unclear whether music lessons influence academic self-concept or whether academic self-concept influences the likelihood of taking music lessons. To sort out the matter of causality, we investigated the influence of an extended music curriculum (EMC) at school on academic self-concept longitudinally. We compared the academic self-concept of children between 9 and 11 years of age before they started the EMC and after a year of EMC and compared it to the academic self-concept of children not attending EMC. We tested 30 children (11 male). Thirteen of them participated in the EMC and 17 did not participate. We measured academic self-concept and confounding variables such as gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, musical aptitude, and motivation. Children with and without EMC did not differ in confounding variables except for nonmusical out-of-school activities. Hence, the amount of nonmusical out-of-school activities was controlled in further analyses. We found an influence of EMC on academic self-concept. Children attending EMC reported significantly higher academic self-concepts after a year of participation than children not attending EMC.

Keywords

academic self-concept children conative transfer extended music curriculum music lessons

The impact of music lessons on cognitive abilities has inspired research aspirations for many years. The associations between music lessons and specific cognitive abilities such as language (Moreno et al., 2009) or memory (Degé, Wehrum, Stark, & Schwarzer, 2011), as well as the associations between music lessons and global cognitive abilities such as IQ (Schellenberg, 2004, 2006) or executive functions (Degé, Kubicek, & Schwarzer, 2011) have been the focus of numerous studies. Aside from these cognitive transfer effects, conative transfer effects have recently also become of more interest for scientific investigations. Conative transfer refers to effects of music lessons on motivational, personality, and self-concept variables (Vitouch, Bischof, & Wieser, 2009). Indeed, music lessons have been associated with personality variables (Corrigall & Schellenberg, 2015; Corrigall, Schellenberg & Misura, 2013) and self-esteem (Costa-Giomi, 2004; Rickard et al., 2012). For a domain specific self-concept, the academic self-concept, an association with music lessons could be demonstrated (Degé et al., 2014). However, the correlational design of this study precludes inferences of causation. Thus, it remains unclear whether taking music lessons influences academic self-concept or whether a high academic self-concept influences the likelihood of taking music lessons. Therefore, our study investigated the association between music lessons and academic self-concept longitudinally to solve the matter of causation. We tested the academic self-concept of 9- to 11-year-old children at the beginning of an extended music curriculum (EMC) at school and after one year of participation and compared it to the academic self-concept of children not attending EMC. Since we tested self-selected groups, we controlled for gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, musical aptitude, and motivation (motivation for schooling) to rule out selection effects and schooling effects as explanation for our results.

Music lessons and cognitive abilities

As mentioned above, positive associations between music lessons and a variety of specific cognitive abilities have been established: music training was linked to language-related abilities such as phonemic awareness (Degé & Schwarzer, 2011; Gromko, 2005), reading (Moreno et al., 2009), and vocabulary (Moreno et al., 2011). Moreover, music lessons were associated with several aspects of memory such as verbal memory (Franklin et al., 2008), visual memory (Jakobson, Lewycky, Kilgour, & Stoesz, 2008), and working memory (Lee, Lu, & Ko, 2007). Furthermore, music lessons were related to mathematical abilities (Vaughn, 2000), spatial abilities (Hetland, 2000), and fine motor skills (Costa-Giomi, 2005). Although each of these findings in itself provides evidence for specific associations, considering them as a whole indicates that associations between music lessons and cognitive abilities are rather general. Because nonmusical benefits of music lessons extended across a wide variety of specific cognitive abilities, Schellenberg (2009) postulated that music lessons were associated with a domain-general cognitive ability that might explain the emergence of all the specific links. IQ is possibly such a domain-general cognitive ability. Therefore, Schellenberg (2004, 2006) investigated the relationship between music lessons and IQ in two comprehensive studies. In a large correlational study he showed in a sample of 6- to 11-year-old children that music lessons were positively associated with IQ even when SES was held constant. In addition, for university students their amount of music lessons in childhood was positively associated with IQ in adulthood. This association also remained reliable when SES was controlled (Schellenberg, 2006). Unfortunately, this approach does not allow us to draw inferences about causation. However, Schellenberg (2004) established with an experiment that music lessons have an impact on IQ. Children were randomly assigned to music lessons (singing and keyboard), a control group that received drama lessons, and a control group that did not receive lessons. Both music groups had significantly higher increases in IQ from pre- to post-test than the two control groups. Apart from IQ, it was found that other domain-general abilities – executive functions – were linked to music lessons. In a correlational study with 9- to 12-year-old children it was demonstrated that several executive functions (set shifting, selective attention, inhibition, fluency, and planning) were positively associated with amount of music lessons (Degé, Kubicek, et al., 2011). In another study with the same age group, Schellenberg (2011) was only able to find an association between music lessons and working memory; none of the other measured executive functions (inhibition, set shifting, fluency, planning) were related to music lessons. However, in a training study, music listening training significantly enhanced inhibition scores of children in the music group compared to children in a visual arts group (Moreno et al., 2011). Taken together, the abovementioned studies provide evidence of associations between music lessons and domain-general cognitive abilities (i.e., IQ and executive functions). These results can be interpreted as supporting evidence for the notion by Schellenberg (2009) that all specific cognitive transfer effects could be explained by a transfer effect of music lessons on domain-general cognitive abilities.

Above and beyond these cognitive transfer effects, conative transfer effects of music lessons have also been discussed. Conative transfer effects of music lessons can be considered as distinct from the cognitive transfer effects (Vitouch et al., 2009). Conative transfer refers to motivational variables, personal styles, or personality variables that influence one’s approach to different tasks (Vitouch et al., 2009). Studies that investigated conative transfer effects revealed inconsistent results. For example, for social competence no transfer effects could be demonstrated, but for motivational and for some self-related variables associations with music lessons have been established. One of these self-related variables is the academic self-concept. The academic self-concept refers to the cognitive representation and appraisal of a person’s academic abilities and is linked to academic achievement. Although the focus of our study is on music lessons and academic self-concept, we will give a broader overview regarding conative transfer effects of music lessons.

Music lessons, social competence, motivational variables, personality, self-esteem, and self-concept

Regarding social competence, in the previously mentioned experiment by Schellenberg (2004) no positive influence of music lessons on adaptive social behavior was found. Quite contrary to expectations, it was not the music lessons, but the drama lessons (which one of the control groups received), which enhanced social competence. Also, in the formerly mentioned correlational study by Schellenberg (2006), no significant positive association between music lessons and social competence, or social adjustment was observed. Furthermore, in a study that investigated the association between emotion comprehension and music lessons, an advantage of the musically trained children disappeared when IQ was held constant (Schellenberg & Mankarius, 2012). Thus, the observed enhanced emotion comprehension could be explained by superior cognitive abilities in the musically trained group. In sum, music lessons that were nearly always private lessons do not influence social competence. Which comes as no surprise, because hours of solitary practice might not result in more sophisticated or competent group interactions. Instead it might be reasonable to investigate associations between group music lessons and social competence.

With respect to motivational variables such as self-regulated learning, no statistically significant differences between musically trained children, children with sports training, and untrained children were found. In this study the participating children had to acquire a new skill (reading mirror writing) and the groups were similarly successful on this task (Bischof, 2008). However, descriptively, the children with music training did show the largest performance gains after self-regulated learning. Nevertheless, this descriptive difference in favor of the musically trained children failed to reach significance. One reason for that might have been the small sample size that limited the power of statistical tests. For perseverance, another motivational variable, an advantage of musically trained children was found. In this study 3- to 5-year-old children were tested with a nonmusical perseverance task. The musically trained children spent more time on this task than children who were trained in creative movement and children in a preschool group (Scott, 1992). In sum, music lessons seem to be associated with motivational variables like perseverance and probably also with self-regulated learning. However, for self-regulated learning no statistically significant effect could be established.

Concerning personality and self-esteem inconsistent results have been reported: personality seems to be linked to music lessons, whereas self-esteem is probably not related to music lessons. Regarding personality, a study with 10- to 12-year-old children revealed that the amount of music lessons was significantly positively correlated with openness-to-experience and conscientiousness (Corrigall et al., 2013). In adults, however, only the personality dimension openness-to-experience was significantly associated with amount of music lessons. Interestingly, for younger children (7- to 9-year-olds) just beginning music lessons, parent’s openness-to-experience score was significantly correlated with duration of children’s music training. For the 7- to 9-year-old children the authors found that the higher a child scored on agreeableness the longer they participated in music lessons (Corrigall & Schellenberg, 2015). However, both studies were correlational in nature, therefore it remains unclear whether music lessons influence personality or whether personality influences the likelihood of taking music lessons. The methodological approach applied by Corrigall and colleagues (2015; 2013), however, implies that it is rather the personality factor (e.g., openness-to-experience) that influences the likelihood of taking music lessons and probably also continuing lessons over a longer period of time than music lessons influencing personality. A twin study regarding personality and music practice showed that openness-to-experience is related to music practice (Butkovic, Ullén, & Mosing, 2015). The authors assume that shared genes influence, on the one hand, a person’s proneness to music practice, and on the other hand the person’s predisposition to a higher openness-to-experience score. With respect to self-esteem, a three-year longitudinal study found marginal significant differences in self-esteem between 9-year-old children receiving piano instruction and children in a control group (Costa-Giomi, 2004). The self-esteem scores of the musically trained children increased over the years, whereas the self-esteem scores of the untrained children did not improve. Moreover, a study with primary school children found an effect of an extended music curriculum at school on self-esteem (Rickard et al., 2012). In this study a younger cohort of children in grade 1 was tested as well as an older cohort of children that attended grade 3. For both cohorts the influence of music lessons on a global measure of self-esteem was found. In addition, in the older cohort an effect of music lessons on academic self-esteem was revealed. In sum, personality, in particular openness-to-experience, is associated with music lessons and music lessons seem to have no impact on self-esteem.

Concerning self-concept, a study that investigated the relationship between academic self-concept and music lessons found a positive association between them (Degé et al., 2014). In a sample of 12- to 14-year-old children the amount of music lessons was positively associated with the reported academic self-concept. This association remained significant even when possible confounding variables like socioeconomic status (i.e., parents’ education), gender, nonmusical out-of-school activities, grade, and IQ were statistically controlled. However, there are two issues concerning this study. First, the role of motivation was not investigated. It might be possible that motivation is a confounding variable that is associated with taking music lessons and high academic self-concept and, in turn, may explain the revealed association between them. Second, due to the correlational nature of this study the direction of causation still needs to be established. It is unclear whether music lessons influence academic self-concept or whether children with high academic self-concept tend to take music lessons.

Academic self-concept

Academic self-concept describes the cognitive representation and appraisal of one’s own abilities in academic performance (Dickhauser, Schöne, Spinath, & Stiensmeier-Pelster, 2002). Experiences with and evaluations by significant others (i.e., parents, peers, and teachers), reinforcements, and attributions for one’s behavior form the academic self-concept (Marsh & O’Mara, 2008; Shavelson, Hubner, & Stanton, 1976). Academic self-concept and academic achievement have a reciprocal relation: as postulated by the reciprocal effects model (Marsh & O’Mara, 2008), an improvement in academic self-concept results in better performance and improvement in performance will enhance academic self-concept (Marsh & Craven, 2006). Self-concept has been shown to be multidimensional (Marsh & Craven, 2006). This implies that children can have different self-concepts in such different areas as physical abilities or academic abilities (Marsh & Craven, 1997). These different self-concepts seem to be ordered hierarchically, and the academic self-concept seems to consist of two higher-order academic factors: the math/academic factor and the verbal/academic factor (Marsh & Craven, 1997). Moreover, associations between academic self-concept and academic achievement were specific to particular school subjects and several dimensions of self-concept were quite distinct (Marsh & Craven, 2006), which provides additional support for the multidimensionality of academic self-concept (Marsh, 1992). In our longitudinal study we assessed the global academic self-concept, which is not as content-specific as a math or verbal academic self-concept. We decided to measure the global academic self-concept because in earlier studies music lessons were associated with an average school grade (Schellenberg, 2006) and with achievement in all school subjects except for sports (Wetter, Koerner, & Schwaninger, 2009).

We expected an influence of music lessons on academic self-concept, because music lessons are a school-like activity that is highly adaptive to students’ abilities. Thus, students experience the value of practice for successful performance, which might have an impact on a student’s behavior in other learning situations. This, in turn, might enhance learning outcomes and changes students’ attitude toward their own abilities in school and in school-like achievement situations (i.e., academic self-concept).

Objectives of the study

Existing studies point toward associations between music lessons and cognitive abilities (i.e., cognitive transfer) as well as associations between music lessons and motivational or personality variables (i.e., conative transfer). Initial studies revealed conative transfer effects for motivation (Bischof, 2008), perseverance (Scott, 1992), self-esteem (Costa-Giomi, 2004; Rickard et al., 2012), and personality (Corrigall & Schellenberg, 2015; Corrigall et al., 2013). Regarding self-concept, a study reported an association between music lessons and academic self-concept (Degé et al., 2014). Because this study was correlational, the direction of causation is unclear. Furthermore, in this study the role of motivation was not investigated, although motivation might contribute to the association between music lessons and academic self-concept. Therefore, the aim of the current study was to examine the impact of music training on academic self-concept longitudinally to solve the matter of causation. We tested 9- to 11-year-old children at the beginning of an extended music curriculum at school and after one year of participation and compared it to the academic self-concept of children not attending EMC. Since we investigated self-selected groups, we controlled for possible confounding variables such as gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, and musical aptitude. Additionally, motivation for school was assessed and statistically controlled to rule out motivational differences between children with and without EMC as explanation for potential differences in academic self-concept.

Method

Participants

The sample consisted of 30 (11 male, 19 female) participants. At the beginning of the study the children ranged in age from 9 to 11 years (M = 10 years; 9 months, SD = 4.6 months). Children were recruited from one secondary school that offered the extended music curriculum (EMC). All children were attending the same track and grade. Children participating and not participating in the EMC were taught together in classes. This reduced the possibility of preferential treatment or Hawthorne effects when studying children with EMC. Of the entire sample, 13 children participated in the EMC and 17 did not participate in the extended music curriculum (nEMC). All children regardless of participating or not participating in the extended music curriculum received one music lesson per week at school. The children participating in EMC had two additional music lessons in school. One of these lessons was devoted to playing on their instruments together with the other EMC participants. Additionally, children attending EMC participated in the school choir and/or in the school orchestra for at least two or up to four hours per week. In addition to the music lessons at school, children attending EMC received music instruction on at least one instrument (no private vocal lessons included) that took place after class. At the beginning of the study all children were involved in organized nonmusical out-of-school activities for at least 6 months. Such activities were, for example, soccer, riding lessons, or participating in the voluntary fire brigade.

The sample showed some diversity with respect to parents’ education: 39.1% of the children had parents with no university degree, 30.4% had one parent with a university degree, and 30.4% had two parents with a university degree. Parents’ education was not significantly correlated with EMC participation (r = .19, p = .39).

Material

We measured possible differences between the children attending EMC and children not attending EMC in confounding variables (gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, musical aptitude, motivation), and we assessed the academic self-concept of each child.

We applied a sociodemographic questionnaire that asked for gender, age, socioeconomic status (SES), and organized nonmusical out-of-school activities. We used mothers’ and fathers’ education as proxy for SES. It was assessed as a dichotomous variable with 0 for “no university education” and 1 for “university education,” and then collapsed into a single variable (0, 1, or 2 parents with a university degree) for the analyses. In addition, parents provided details about their child’s history of nonmusical out-of-school activities. An overall score of nonmusical out-of-school activities in months was computed for each child on the basis of this information. If a child participated in more than one activity at a time, the months for each activity were summed.

IQ was measured with a short form of the German version of the Wechsler Intelligence Scale for Children (HAWIK-III) (Tewes, Rossmann, & Schallberger, 2000), which consisted of two verbal (vocabulary, information) and two performance (picture arrangement, block design) subtests. The split-half reliability for the different subtests were: vocabulary r = .88, information r = .85, picture arrangement r = .75, and block design r = .88. The correlation between the subtests and the general IQ score were: vocabulary r = .60, information r = .60, picture arrangement r = .53, and block design r = .56. An estimate of full-scale IQ was computed based on these four subtests according to the formula provided by Schallberger (2005). The short version explains R² = .92 of the variance in full-scale IQ. This estimate of full-scale IQ was used in further statistical analyses.

Musical aptitude was assessed using the Advanced Measures of Music Audiation (Gordon, 1989). The Advanced Measures of Music Audiation is a CD-based music aptitude test that can be administered in a group setting. The CD-recording comprises the instruction for test taking, three practice exercises and 30 test items. Each test item consists of two music phrases, a standard and a probe. The probe could be identical to the standard or represent a change in either tone or rhythm. Participants were asked to listen attentively to each item and subsequently deliver a same/different judgement. Thus, participants had to respond to tonal and rhythm aspects of the music phrases at the same time. Test administration, excluding distribution and collection of the answer sheets, required less than 20 minutes. The test-retest reliability for the subtest scores and the total score ranges between r = .80 and .87. The validity ranges between r = .43 and .78.

Motivation was assessed with a questionnaire (Skalen zur Erfassung der Lern-und Leistungsmotivation; Spinath, Stiensmeier-Pelster, Schöne & Dickhäuser, 2002). This questionnaire assesses four school/learning related motivational aspects: learning goals, approach performance goals, avoid performance goals, and work avoidance. The dimension “learning goals” refers to a person’s motivation to learn something (new) at school independent of performance. The dimension “approach performance goals” describes a person’s willingness to work on viable tasks to demonstrate competence/good performance at school. The dimension “avoid performance goals” describes a person’s tendency to avoid difficult tasks and to hide missing competence/knowledge as well as possible. The dimension “motivation to avoid work” indicates how much someone tries to avoid difficult tasks at school. The dimension learning goals consists of 7 items, whereas the other three dimensions consist of 8 items. Each item has a five-stage response format. For each dimension the response scores of the belonging items were summed. The split-half reliability for the different motivational aspects were: learning goals r = .73, approach performance goals r = .74, avoid performance goals r = .77, and motivation to avoid work r = .78. Construct validity for learning goals was r = .54, for approach performance goals r = .53, for avoid performance goals r = .31, and for motivation to avoid work r = .41.

Academic self-concept was assessed as a primary dependent variable. It was measured with a questionnaire (Skalen zur Erfassung des Schulischen Selbstkonzeptes [SESSKO]; Schöne, Dickhäuser, Spinath, & Stiensmeier-Pelster, 2002) that provided the opportunity to assess a global estimate of academic self-concept and three specific frame-of-reference estimates of academic self-concept (individual, social, and criterion based). Only the global measure of academic self-concept was used because we did not have a hypothesis that music lessons would have different effects on different frames of reference. The questionnaire consisted of 22 items. Each item offered a five-stage response format. All items contained a statement and two poles of a semantic differential. The measure for the global academic self-concept consisted of five items. The global academic self-concept score provided a general estimate of the self-reported academic self-concept. For example, an item assessing the overall estimate of school ability on a five-stage scale had at one end the statement: “at school the different tasks are difficult for me,” and at the other end the statement: “at school the different tasks are easy for me.” Responses were scored from 1 to 5, and the score on each item was summed to build a global score for academic self-concept. The questionnaire allows transfer of these scores into t-scores. This transfer was performed for further analysis. The split-half reliability r =.84 and the criterion validity r =. 66 of the academic self-concept scale are satisfactory.

Procedure

Children were tested twice. T0 was at the beginning of the EMC and T1 was after a year of EMC or no EMC, respectively. The procedure was exactly the same for T0 and T1. The testing procedure consisted of an individual session, a group session, and the completion of the sociodemographic questionnaire. Prior to the test sessions parents received the sociodemographic questionnaire either via mail or email and completed it. Intelligence was assessed in the individual test session. It took approximately 30–40 minutes to administer the short version of the IQ test. In the group test session musical aptitude, motivation, and academic self-concept were assessed. The administration of the musical aptitude test, excluding distribution and collection of the answer sheets, required less than 20 minutes. Children worked for approximately 15 minutes on the motivation questionnaire and the academic self-concept questionnaire. Both questionnaires were administered before the children worked on the music aptitude test to avoid any influence of the test on children’s estimate of academic self-concept. Children were tested by female assistants trained in administering the short version of the IQ test, the musical aptitude test, the motivation questionnaire, and the academic self-concept questionnaire. The assistants were not aware of the EMC participation of the children. Hence, they did not know if the child they tested was participating in the EMC or not participating in the EMC. All children received a gift certificate as a means of thanking them for their participation.

Results

Preliminary analyses

We calculated Kolmogorov–Smirnov tests to check whether our variables were normally distributed. In case of a significant deviance from normal distribution we complemented our calculations with the most appropriate non-parametrical tests. One measurement at T0 or T1 deviated significantly from normal distribution for the variables: learning goals, avoid performance goals, motivation to avoid work, and academic self-concept. See Table 1 for the test statistics and the kurtosis and skewness values of these variables.

Table 1.

Skewness, kurtosis and results of the Kolmogorov–Smirnov test for age, nonmusical out-of-school activities, IQ, music aptitude, motivation measures, and academic self-concept.

	skewness	kurtosis	K–S	df	p
age	−.46	−.52	.16	30	.17
out–of–school activities T0	.62	−.11	.11	29	.20
out–of–school activities T1	1.08	.87	.16	29	.07
IQ T0	.42	1.74	.13	30	.19
IQ T1	.30	−1.02	.12	30	.20
music aptitude T0	.22	1.37	.16	30	.06
music aptitude T1	.21	−.51	.07	30	.20
learning goals T0	−1.68	3.5	.19	29	.01*
learning goals T1	−1.56	4.33	.12	29	.20
approach performance goals T0	.19	−.91	.09	29	.20
approach performance goals T1	.32	−.29	.11	29	.20
avoid performance goals T0	.23	.41	.14	29	.14
avoid performance goals T1	.12	−1.39	.19	29	.01*
motivation to avoid work T0	.69	.28	.12	29	.20
motivation to avoid work T1	.47	−.66	.17	29	.04*
academic self–concept T0	−.65	.23	.19	30	.01*
academic self–concept T1	.36	.66	.10	30	.20

Note. *p < .05.

To ensure the comparability of children with and without EMC, we controlled for possible confounding effects of gender, age, SES, nonmusical out-of-school activities, IQ, music aptitude, and motivation.

The female to male ratio between the groups did not differ significantly, χ² = (1, n = 30) 3.44, p = .56. In the group of children attending EMC 30.77% were male, and in the group not attending EMC 41.18% were male, see Table 2.

Table 2.

Distribution of males and females in the group of children attending and not attending EMC.

	Male	Female
EMC	4	9
nEMC	7	10

Regarding mean age, the children with EMC (M = 129.31 months, SD = 3.84) and without EMC (M = 129.12 months, SD = 5.29) were comparable, t(28) = -0.11, p = .91. Children with and without EMC were also comparable in their SES, χ² = (1, n = 23) 2.84, p = .24, see Table 3 for details. All in all seven parents did not provide details about their education (5 EMC, 2 nEMC).

Table 3.

Distribution of parents’ education as a measure of socioeconomic status within the group of children attending EMC (EMC) and not attending EMC (nEMC).

	0 parent	1 parent	2 parents
EMC	3	1	4
nEMC	6	6	3

Concerning nonmusical out-of-school activities, months of actually practiced activities were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found a marginal significant main effect of condition, F(1, 27) = 3.81, p = .06, η² = 0.12, see Table 4 for means and standard deviations. Children with EMC as well as children without EMC showed increases in amount of nonmusical out-of-school activities from T0 to T1. Additionally, we found a significant main effect of group, F(1, 27) = 8.95, p = .01, η² = 0.25. Children with and without EMC differed significantly in their amount of nonmusical out-of-school activities. Children with EMC reported a higher amount of nonmusical out-of-school activities than children without EMC. Also, the group × condition interaction was significant, F(1, 27) = 4.37, p = .05, η² = 0.14. Since the values for T0 and T1 of nonmusical out-of-school activities were significantly correlated (r = .78, p = .01), we additionally calculated t-tests that compared the groups at each time point. At T0 the children with EMC reported significantly more nonmusical out-of-school activities, t(17.45) = -2.18, p = .04. The same holds true for T1, t(18.03) = -3.16, p = .01.

Table 4.

Means and standard deviations (in parentheses) for amount of nonmusical out-of-school activities at T0 and T1 for children attending EMC (EMC) and children not attending EMC (nEMC).

	T0	T1
EMC	150.54 (94.24)	194.15 (105.21)
nEMC	92.13 (49.38)	90.63 (59.30)

Estimated full-scale IQ scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found a significant main effect of condition, F(1, 28) = 19.94, p <.001, η² = 0.42. Children with EMC (M_T0 = 109.8, SD_T0 = 15.55, M_T1 = 117.36, SD_T1 = 16.66) as well as children without EMC (M_T0 = 106.6, SD_T0 = 12.04, M_T1 = 117.67, SD_T1 = 12.87) showed increases in full-scale IQ from T0 to T1. We found no significant main effect of group, F(1, 28) = 0.09, p = .76. Children with and without EMC did not differ in their full-scale IQ. Also, the group × condition interaction was not significant, F(1, 28) = 0.71, p = .41.

Musical aptitude scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We revealed no significant main effect of condition, F(1, 28) = 3.02, p = .09. The participants had comparable musical aptitude scores at T0 and T1. Additionally, we found no significant main effect of group, F(1, 28) = 0.19, p = .67. Children with and without EMC did not differ in their musical aptitude scores. Also, the group × condition interaction was not significant, F(1, 28) = 2.01, p = .66, see Table 5 for means and standard deviations. Also the values of music aptitude for T0 and T1 were significantly correlated (r = .38, p = .04). Hence, we calculated t-tests that compared the groups at each time point. At T0 the children with and without EMC did not differ significantly in musical aptitude, t(28) = -0.11, p = .91. The same holds true for T1, t(28) = -0.61, p = .55.

Table 5.

Means and standard deviations (in parentheses) for musical aptitude at T0 and T1 for children attending EMC (EMC) and children not attending EMC (nEMC).

	T0	T1
EMC	46.69 (4.92)	50.38 (6.28)
nEMC	46.35 (9.94)	48.53 (9.52)

See Table 6 for means and standard deviations of all motivation measures. All four different motivation scores were entered into an ANOVA with repeated measures. Learning goals scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found no significant main effect of condition, F(1, 27) = 2.82, p = .11. The children reported comparable learning goal orientation at T0 and T1. Also, we found no significant main effect of group, F(1, 27) = 1.93, p = .18. Children with and without EMC did not differ in their reported learning goal orientation. The group × condition interaction, as well, was not significant, F(1, 27) = 1.66, p = .21. Since the assumption of normal distribution was violated for learning goals at T0, we calculated an additional Mann-Whitney-U test (for T0: U = 64, p = .053 and for T1: U = 94.5 p = .68).

Table 6.

Means and standard deviations (in parentheses) for motivation variables at T0 and T1 for children attending EMC (EMC) and children not attending EMC (nEMC).

	learning goals T0	learning goals T1	approach performance goals T0	approach performance goals T1	avoid performance goals T0	avoid performance goals T1	motivation to avoid work T0	motivation to avoid work T1
EMC	36.23	33.38	24.69	23.08	21.62	20.38	21.77	19.15
EMC	(3.44)	(5.50)	(5.17)	(4.73)	(6.04)	(5.99)	(8.37)	(7.02)
nEMC	32.59	31.94	24.94	21.75	22.35	19.19	22.71	20.63
nEMC	(5.93)	(7.09)	(4.10)	(4.31)	(5.16)	(5.81)	(7.26)	(6.88)

Approach performance goals scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found a significant main effect of condition, F(1, 27) = 7.27, p = .01, η² = 0.21. Children with EMC (M_T0 = 24.69, SD_T0 = 5.17, M_T1 = 23.08, SD_T1 = 4.73) as well as children without EMC (M_T0 = 24.94, SD_T0 = 4.1, M_T1 = 21.75, SD_T1 = 4.3) showed decreases in their approach performance goals from T0 to T1. We found no significant main effect of group, F(1, 27) = 0.8, p = .79. Children with and without EMC did not differ in their reported approach performance goals. Additionally, the group × condition interaction was not significant, F(1, 27) = 0.99, p = .33. Avoid performance goals scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found no significant main effect of condition, F(1, 27) = 3.27, p = .08. The children reported comparable avoid performance goals at T0 and T1. Also, we found no significant main effect of group, F(1, 27) = 0.04, p = .84. Children with and without EMC did not differ in their avoid performance goals scores. The group × condition interaction, as well, was not significant, F(1, 27) = 0.53, p = .48. Also, the avoid performance goals scores were tested with a non-parametric test, because at T1 the scores violated the assumption of normal distribution: for T0: U = 101, p = .69 and for T1: U = 96.5 p = .74).

Motivation to avoid work scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of variance (ANOVA) with repeated measures on the last factor. We found no significant main effect of condition, F(1, 27) = 2.29, p = .14. The children reported comparable motivation to avoid work at T0 and T1. Also, we found no significant main effect of group, F(1, 27) = 0.14, p = .71. Children with and without EMC did not differ in their reported motivation to avoid work. The group × condition interaction was also not significant, F(1, 27) = 0.19, p = .66. The additional non-parametric analyses due to the violation of the assumption of normal distribution did not reveal any significant differences between children with and without EMC at T0, U = 105, p = .82, or T1, U = 92, p = .60.

Based on these preliminary analyses we did not consider gender, age, SES, IQ, musical aptitude, and motivation as covariates in the principal analyses. For the nonmusical out-of-school activities we calculated a difference score T1–T0 and used this as covariate in the principal analyses.

Principal analyses

Although the academic self-concept variable was not normally distributed at T0, we decided to run an ANCOVA. On the one hand this gave us the possibility to control influences of nonmusical out-of-school activities and on the other hand Analyses of variance are quiet robust procedures even when the assumption of normal distribution is violated (Field, 2005). However, for the further statistical analyses we used non-parametrical tests. Academic self-concept scores were entered into a 2 (group: EMC vs. nEMC) × 2 (condition: T0 vs. T1) analysis of covariance (ANCOVA) with repeated measures on the last factor and difference scores of nonmusical out-of-school activities as a covariate. No significant condition × covariate interaction, F(1, 26) = 2.29, p = .14, was found. The analysis showed no significant main effect of condition, F(1, 26) = 0.53, p = .47. The children reported comparable academic self-concepts at T0 and T1. Additionally, we found no significant main effect of group, F(1, 26) = 1.41, p = .25. Children with and without EMC did not differ in their reported academic self-concept. However, the group × condition interaction was significant, F(1, 26) = 4.27, p = .049, η² = 0.14. See Figure 1. In further analyses this effect was inspected more closely. Due to the violation of normal distribution we applied non-parametric tests for further analyses. The effects of condition were inspected with Mann-Whitney-U tests. At T0 there was no difference between children with and without EMC in reported academic self-concepts, U = 106, p = .85. However, at T1 children with and without EMC differed in their reported academic self-concepts, U = 65, p = .05. Children attending EMC reported higher academic self-concepts than children without EMC. The effects of group were further investigated with the Wilcoxon signed-rank test. Children not attending EMC showed no significant change from T0 to T1 in their academic self-concept, T = -1.24, p = .21. Numerically they rather showed a decrease from T0 (M = 55.24, SD = 10.16) to T1 (M = 51.41, SD = 8.34) in the reported academic self-concept scores. In contrast, children attending EMC showed a marginal significant increase from T0 (M = 55.54, SD = 9.86) to T1 (M = 59.15, SD = 10.34) in their academic self-concept, T = -1.83, p = .068.

Discussion

We investigated the impact of music lessons on academic self-concept. In a longitudinal study the academic self-concept of 9- to 11-year-old children was tested at the beginning of an extended music curriculum at school and after one year of participation and compared to the academic self-concept of children not attending EMC. Because we tested self-selected groups we controlled for gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, musical aptitude, and motivation for school, to rule out systematic differences between children with and without EMC in confounding variables as explanations for our results. With this longitudinal approach, we found that a year of EMC has probably enhanced academic self-concept. At T0, the beginning of the EMC, children with and without EMC did not differ in academic self-concept. A year later, however, children with EMC reported a significantly higher academic self-concept than children without EMC.

The influence of music lessons on academic self-concept

What is it about music lessons that might influence academic self-concept? Music lessons in general are a school-like activity (Schellenberg, 2006). Especially in this study music instruction was embedded in a school curriculum, the extended music curriculum. These points may contribute to an influence of music lessons on academic self-concept, because they might promote transfer from music lessons to school settings. Moreover, music lessons normally yield positive feedback, because they are mostly taught individually or in small groups. This gives the teacher the opportunity to assign viable homework and foster learning at an individual pace. Furthermore, music lessons enable the learner to experience that a particular ability is malleable. Usually a piece of music is mastered after a bit of practice and practice is rewarded with progress. Such an experience might foster the learning of a key concept for every kind of learning, not only in music, but also in other domains (e.g., school). This more mastery-oriented approach of learning situations in school might result in better school performance and may, thus, also influence academic self-concept via this route.

Music lessons and conative transfer effects

Our results complement already existing findings in the conative transfer literature. In accordance with the correlational results by Degé and colleagues (2014), we found an emerging association between music lessons and academic self-concept. Both studies tested children in a similar age range: 10- to 12-year-old children in the correlational study and 9- to 11-year-old children in our study (they were 10- to 12-year-old at T1). Also, the applied measure of academic self-concept was the same. Hence, our study replicated former results. However, we broadened the knowledge about music lessons and academic self-concept significantly. With our longitudinal design we were able to demonstrate that music lessons have an impact on academic self-concept and thereby shed light on the causal relationship between them. It is important to note that due to our small sample size and the self-selected groups our result needs replication to gain generalizability. We inspected the contribution of motivation to the association between music lessons and academic self-concept. We found that the association is not explained by motivation for school. It is important to keep in mind that motivation to learn an instrument or influences of general motivation have not been controlled yet. Therefore, such influences cannot be ruled out. It is possible, for example, that motivation to participate in after-school activities in general may have been higher in the EMC group. The differences in participation in such activities may indicate this. Hence, it is important for future research to assess further aspects of motivation.

Costa-Giomi (2004) found a marginal significant impact of music lessons on self-esteem in 9-year-old children. In more detailed analyses the author found that academic self-esteem was the subcomponent of the self-esteem measure that was significantly influenced by music lessons. Similarly, Rickard and colleagues (2012) revealed that in primary school children a global measure of self-esteem as well as a measure of academic self-esteem was positively influenced by music lessons. These findings go along well with our results. Both suggest that music lessons might have an effect on the self, especially when it comes to the academic domain. When integrating these results it seems that music lessons influenced academic self-esteem, a more affective measure of a child’s estimation of abilities (Costa-Giomi, 2004; Rickard et al., 2012), as well as academic self-concept (i.e., the cognitive appraisal of a person’s abilities). Like our study the Rickard study investigated the influence of a school-based music intervention. Thus, both studies revealed a positive influence of a school music program on measures of self.

Taken together, conative transfer of music lessons has been demonstrated for several different variables such as self-regulated learning (Bischof, 2008), perseverance (Scott, 1992), personality (Corrigall & Schellenberg, 2015; Corrigall et al., 2013), self-esteem (Costa-Giomi, 2004), and academic self-concept (Degé et al., 2014). Notably, our study showed that music lessons might have the potential to enhance academic self-concept. Interestingly, studies were not able to demonstrate that music lessons were linked to social competence (Schellenberg, 2004, 2006). This null finding might be a consequence of the kind of lessons. So far, studies have investigated the influence of individual music lessons on social competence. However, it might be more promising to investigate the impact of group music lessons on social competence.

Music lessons and cognitive transfer effects

Regarding intelligence, one of our control variables, we revealed an increase between T0 and T1 for the children with and without EMC. This increase might represent an effect of development, education, or a retest effect and no specific effect of music training. This finding might seem contradictory to findings from Schellenberg (2004). However, the music lessons we investigated differ with respect to age of onset from the music lessons in Schellenberg’s study. The children with EMC started music training later in life than the children in the study by Schellenberg (2004). The extended music curriculum starts in secondary school (9- to 11-year-old children), whereas the music training in Schellenberg’s study addressed 6-year-old children. It is probable that the beginning of music lessons around six years of age might result in different outcomes than secondary school music training. In line with this assumption are, for example, findings of a two-year longitudinal study with secondary school children that also did not reveal an effect of music lessons on IQ (Degé et al., 2011).

Limitations and future directions

Selection effects due to testing already existing groups might limit our results; participants were not assigned randomly to the music lessons. However, we accounted for possible selection effects by measuring the control variables (gender, age, socioeconomic status, organized nonmusical out-of-school activities, IQ, musical aptitude, and motivation for school). This approach should have ruled out selection effects or alternative explanations for the positive effect of music lessons on academic self-concept. In addition, the groups did not differ at T0 in nearly all of the control variables (except for nonmusical out-of-school activities) and the dependent variable. Another limitation is the small sample size. It limits generalizability. Hence, future research should aim at replicating our results with a larger sample. However, our study is the first to provide evidence of an influence of music lessons on academic self-concept. This can be an important starting point for future research concerning music lessons and academic achievement. Aside from associations between music lessons and cognitive abilities or the relationship between music lessons and motivational or personality variables, it was shown that music lessons are linked to academic achievement (Schellenberg, 2006; Wetter et al., 2009). This association cannot be explained by a pure cognitive transfer effect. Thus, the relation between music lessons and academic achievement is not explained by the influence music lessons have on IQ (Schellenberg, 2006). Thus, a conative transfer effect might contribute to the emergence of this association. Therefore, the most important point for future research is to investigate the contribution of academic self-concept to the association between music lessons and academic achievement.

Footnotes

Acknowledgements

The authors would like to thank the participants and their parents for participating in the study. The study was conducted in full accordance with the Ethical Guidelines of the German Association of Psychologists (DGPs). In accordance with the ethical guidelines mentioned above informed consent was obtained from the parents for each participant.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

Bischof

(2008). Spiegelschriftlesen: Bessere Leistungen bei MusikerInnen durch konativen Transfer und Kontextaktivierung? [Reading mirror writing: Better achievements of musicians due to conative transfer and context effects] (Diploma thesis). Alpen-Adria-Universität, Klagenfurt, Austria.

Butkovic

Ullén

Mosing

M. A.

(2015). Personality related traits as predictors of music practice: Underlying environmental and genetic influences. Personality and Individual Differences, 74, 133–138.

Corrigall

K. A.

Schellenberg

E. G.

(2015). Predicting who takes music lessons: Parent and child characteristics. Frontiers in Psychology, 6, 282.

Corrigall

K. A.

Schellenberg

E. G.

Misura

N. M.

(2013). Music training, cognition, and personality. Frontiers in Psychology, 4(222). doi:10.3389/fpsyg.2013.00222

Costa-Giomi

(2004). Effects of three years of piano instruction on children’s academic achievement, school performance and self-esteem. Psychology of Music, 32(2), 139–152.

Costa-Giomi

(2005). Does music instruction improve fine motor abilities? Neurosciences and Music II: From Perception to Performance, 1060, 262–264.

Degé

Kubicek

Schwarzer

(2011). Music lessons and intelligence: A relation mediated by executive functions. Music Perception, 29(2), 195–201.

Degé

Schwarzer

(2011). The effect of a music program on phonological awareness in preschoolers. Frontiers in Psychology, 2(124). doi:10.3389/fpsyg.2011.00124

Degé

Wehrum

Stark

Schwarzer

(2011). The influence of two years of school music training in secondary school on visual and auditory memory. European Journal of Developmental Psychology, 8(5), 608–623.

10.

Degé

Wehrum

Stark

Schwarzer

(2014). Music lessons and academic self-concept in 12- to 14-year-old children. Musicae Scientiae, 18(2), 203–215.

11.

Dickhauser

Schöne

Spinath

Stiensmeier-Pelster

(2002). The Academic Self Concept Scales: Construction and evaluation of a new instrument. Zeitschrift fur Differentielle und Diagnostische Psychologie, 23(4), 393–405.

12.

Field

(2005). Discovering statistics using SPSS. London, UK: SAGE.

13.

Franklin

M. S.

Moore

K. S.

Yip

C. Y.

Jonides

Rattray

Moher

(2008). The effects of musical training on verbal memory. Psychology of Music, 36(3), 353–365.

14.

Gordon

E. E.

(1989). Advanced Measures of Music Audiation. Chicago: GIA.

15.

Gromko

J. E.

(2005). The effect of music instruction on phonemic awareness in beginning readers. Journal of Research in Music Education, 53(3), 199–209.

16.

Hetland

(2000). Learning to make music enhances spatial reasoning. Journal of Aesthetic Education, 34(3–4), 179–238.

17.

Jakobson

L. S.

Lewycky

S. T.

Kilgour

A. R.

Stoesz

B. M.

(2008). Memory for verbal and visual material in highly trained musicians. Music Perception, 26(1), 41–55.

18.

Lee

(2007). Effects of skill training on working memory capacity. Learning and Instruction, 17, 336–344.

19.

Marsh

H. W.

(1992). The content specificity of relations between academic achievement and academic self-concept. Jounal of Educational Psychology, 84, 35–45.

20.

Marsh

H. W.

Craven

R. G.

(1997). Academic self-concept: Beyond the dustbowl. In Phye

G. D.

(Ed.), Handbook of classroom assessment: Learning, achievement, and adjustment (pp. 131–198). San Diego, CA: Academic Press.

21.

Marsh

H. W.

Craven

R. G.

(2006). Reciprocal effects of self-concept and performance from a multidimensional perspective. Perspectives on Psychological Science, 1(2), 133–163.

22.

Marsh

H. W.

O’Mara

(2008). Reciprocal effects between academic self-concept, self-esteem, achievement, and attainment over seven adolescent years: Unidimensional and multidimensional perspectives of self-concept. Personality and Social Psychology Bulletin, 34(4), 542–552.

23.

Moreno

Bialystok

Barac

Schellenberg

E. G.

Cepeda

N. J.

Chau

(2011). Short-term music training enhances verbal intelligence and executive function. Psychological Science, 22(11), 1425–1433.

24.

Moreno

Marques

Santos

Castro

S. L.

Besson

(2009). Musical training influences linguistic abilities in 8-year-old children: More evidence for brain plasticity. Cerebral Cortex, 19(3), 712–723.

25.

Rickard

N. S.

Appelman

Richard

Murphy

Gill

Bambrick

(2012). Orchestrating life skills: The effect of increased school-based music classes on children’s social competence and self-esteem. International Journal of Music Education, 31(3), 292–309.

26.

Schallberger

(2005). Welches sind die nach statistischen Kriterien besten Kurzformen des HAWIK-III? [Which short versions of the WISC-III are the best ones according to statistical criteria?]. Berichte aus der Abteilung Angewandte Psychologie, 32, 3–29.

27.

Schellenberg

E. G.

(2004). Music lessons enhance IQ. Psychological Science, 15(8), 511–514.

28.

Schellenberg

E. G.

(2006). Long-term positive associations between music lessons and IQ. Journal of Educational Psychology, 98(2), 457–468.

29.

Schellenberg

E. G.

(2009). Musikunterricht, geistige Fähigkeiten und Sozialkompetenz: Schlussfolgerun-gen und Unklarheiten [Music lessons, mental abilities and social skills: Conclusions and uncertainties]. In Bundesministerium für Bildung und Forschung (Ed.), Pauken mit Trompeten [Cramming with trumpets: Can learning strategies, learning motivation and social skills be facilitated by music lessons?] (pp. 114–124). Berlin, Germany: Bundesministerium für Bildung und Forschung.

30.

Schellenberg

E. G.

(2011). Examining the association between music lessons and intelligence. British Journal of Psychology, 102, 283–302.

31.

Schellenberg

E. G.

Mankarius

(2012). Music training and emotion comprehension in childhood. Emotion, 12, 887–891.

32.

Schöne

Dickhäuser

Spinath

Stiensmeier-Pelster

(2002). SESSKO Skalen zur Erfassung des schulischen Selbstkonzeptes [Scales for the measurement of academic self-concept]. Göttingen, Germany: Hogrefe.

33.

Scott

(1992). Attention and perseverance behaviours of preschool-children enrolled in suzuki violin lessons and other activities. Journal of Research in Music Education, 40(3), 225–235.

34.

Shavelson

R. J.

Hubner

J. J.

Stanton

G. C.

(1976). Self-concept: Validation of construct interpretations. Review of Educational Research, 46(3), 407–441.

35.

Spinath

Stiensmeier-Pelster

Schöne

Dickhäuser

(2002). SELLMO-Skalen zur Erfassung der Lern-und Leistungsmotivation [Scales for the measurement of learning and achievement motivation]. Göttingen, Germany: Hogrefe-Verlag.

36.

Tewes

Rossmann

Schallberger

(2000). Hamburg-Wechsler-Intelligenztest für Kinder III. [Wechsler-Intelligence-Scale for children III]. Bern, Switzerland: Hans Huber.

37.

Vaughn

(2000). Music and mathematics: Modest support for the oft-claimed relationship. Journal of Aesthetic Education, 34(3–4), 149–166.

38.

Vitouch

Bischof

Wieser

(2009). Konativer Transfer: vom Know-What zum Know-How? [Conative transfer: from know-what to know-how]. In Bundesministerium für Bildung und Forschung (Ed.), Pauken mit Trompeten [Cramming with trumpets: Can learning strategies, learning motivation and social skills be facilitated by music lessons?] (pp. 137–147). Berlin, Germany: Bundesministerium für Bildung und Forschung.

39.

Wetter

O. E.

Koerner

Schwaninger

(2009). Does musical training improve school performance? Instructional Science, 37(4), 365–374.