Abstract
The effect of background music on learning and academic performance is highly relevant to the everyday lives of millions of students worldwide. To investigate this, we recruited 165 undergraduate students (71 males and 94 females, mean age of 21.87 years) to complete arithmetic, reading comprehension, and word memory tasks while exposed to familiar or unfamiliar, foreign or first language music, and no music. With the task scores as dependent variables, a significant main effect was observed for music familiarity but not for language. Further analysis showed that only the word memory task was affected by music with significantly higher scores in the familiar than unfamiliar music conditions. Despite a slight negative trend, no significant effect was found in the interaction between music and language conditions. Depending on the task, familiarity but not language of music affected learning and task performance when compared to the no music condition.
Studying while playing background music, especially popular vocal music (Teo, 2005), has become a common habit among students (Cockrill, Sullivan, & Norbury, 2011). The concern with whether music might interfere with the learning and performance of students lies in the minds of many. It is commonly perceived that background music negatively affects studying (see, e.g., Lavoie & Collins, 1975; Moreno & Mayer, 2000) especially since music is processed in areas of the brain that also serve many other general functions (Ho, Cheung, & Chan, 2003; Janata, Tillmann, & Bharucha, 2002; Sharps & Pollitt, 1998), where activity in one hemisphere constrains activity in the other (Kinsbourne, 1974). Arising from this is the question of whether the language of lyrics might affect learning and performance given that the effects of lyrics remain inconclusive. For example, Greitemeyer (2009) found that song lyrics elicited prosocial behaviour in the listener, but Pieschl and Fegers (2015) only managed to replicate the findings in aggressive lyrics but not prosocial lyrics.
While music has been found to affect emotional arousal (Gan, Lim, & Haw, 2015; Salimpoor, Benovoy, Larcher, Dagher, & Zatorre, 2011), this has been proposed to be mitigated by familiar music, to which one often becomes habituated (Culbert & Posner, 1960, cited in Wolf & Weiner, 1972). Based on the habituation hypothesis (Wolf & Weiner, 1972), familiar music affected performance positively, making familiarity a better predictor of performance than the volume of the music. Background music is thus less disruptive for tasks requiring minimal central verbal component involvement such as arithmetic (Miller & Schyb, 1989), and is in fact found to be beneficial in reducing anxiety (Barger, 1979; Gan et al., 2015).
To address these different reports on the varying effect of music, and in an Asian context, we investigated the effects of the familiarity and the language of vocal music on reading comprehension, word memory, and arithmetic tasks, to determine if: (1) familiarity of music and language of lyrics will affect learning and task; (2) familiarity of music will affect performance on (a) arithmetic, (b) word memory and (c) reading comprehension tasks; and (3) understanding the language of the song lyrics will affect performance on (a) arithmetic, (b) word memory and (c) reading comprehension tasks. These findings would be of interest to every student, educator and parent.
Method
Study
A pilot study (n = 5) was conducted but revealed a ceiling effect in task performance. Thus the duration for each task was shortened and the rest breaks in-between the tasks were removed for the actual study. The between-participant design of the pilot study was carried over to the actual study to keep the total duration short to prevent fatigue frustration and habituation.
Participants
A total of 165 undergraduates (71 males, 94 females) from James Cook University, Singapore, were recruited. The participants, aged 18 to 30 (M = 21.87, SD = 2.36), were recruited based on their English language competency and Italian language illiteracy. It was made known to participants that study involvement was entirely voluntary and withdrawal could take place at any time during the study without prejudice.
Materials
A total of five conditions were used in this quasi-experimental design: i) familiar English language song; ii) the Italian language version of the song in Condition 1; iii) unfamiliar Italian language song; iv) the English language version of Condition 3, and v) no music. The familiar English language song used was “My Heart Will Go On” sung by Celine Dion, while the Italian version was “Il Mio Cuore Va” sung by Sarah Brightman. The unfamiliar Italian language song used was “Volare” sung by Luciano Pavarotti, while the English version was sung by Engelbert Humperdinck. The dependent variables (DV) were the test materials. The arithmetic task and the comprehension passage were taken from the 2012 Key Stage 2 SAT Mathematics and Reading Comprehension papers. Key Stage 2 was chosen for its wide applicability in the local population across most education levels. To study whether music has an effect on simple learning and performance tasks, other test materials included: a) a list of 20 English words (see supplementary materials) randomly selected from the Toronto Word Pool (Friendly, Franklin, Hoffman, & Rubin, 1982); b) a list of English words containing fifteen words from the first list and five new words associated with the themes of the songs (see supplementary materials); c) a short comprehension passage “At the Sign of the Sugared Plum” by Mary Hooper, taken from the 2012 Key Stage 2 SAT Reading Comprehension Paper (see supplementary materials); d) 10 arithmetic questions taken from the 2012 Key Stage 2 SAT Mathematics Paper A (see supplementary materials); e) an information sheet; f) an informed consent form; and g) a demographics form. Permission to use the 2012 Key Stage 2 SAT Reading Comprehension Paper and Mathematics Paper was obtained.
Procedure
The study was carried out in sessions in a quiet room that could comfortably accommodate five participants and three researchers. The music was played through the public announcement system in the room at a fixed volume from the onset until the completion of the entire procedure. Participants were given the tasks in randomized order to minimize the effect of other factors (e.g., fatigue, habituation or frustration with the music). Upon providing written consent, the demographic forms were collected prior to proceeding to the three tasks.
In the reading comprehension task, participants were given five minutes to read the comprehension passage followed by another five minutes to respond to seven forced-choice questions based on the contents of the passage. In the word memory task, participants were given one minute to memorise the list of twenty randomly selected English words from the Toronto Word Pool prior to removal of the list and provision of a new list consisting of fifteen words from the original list and five new words from the theme of the two songs. Participants were asked to identify and tick against words found in the original list within three minutes. In the arithmetic task, participants were required to complete ten mathematical questions without the aid of a calculator in fifteen minutes.
Statistical analysis was performed using IBM SPSS 20.0. Three scores were computed for each participant: arithmetic score, reading comprehension score and word memory score. A two-way between-groups multivariate analysis of variance (MANOVA) was performed to investigate task score differences in the presence of different song and language conditions for the study and control (no music) groups. A one-way between-groups analysis of variance (ANOVA) and independent samples t-test were used to further analyse the main effect found using the MANOVA. Regression was used to assess the ability of the predictor gender to affect task scores.
Ethical considerations
Ethical approval (approval number H5138) was given by James Cook University Human Research Ethics Committee. The experiment was deemed to be a low risk study.
Results
Assumption testing
Preliminary assumption testing was first conducted. The assumptions of normality, linearity, homogeneity of variance–covariance matrices were checked for MANOVA while normality, linearity, multicollinearity and singularity, homoscedasticity and independence of residuals were checked for regression. Although the univariate normality assumption was violated, there were at least 33 cases in each cell, more than the recommended “at least twenty” to ensure robustness according to Tabachnick and Fidell (2007). Hence the analysis continued while noting the assumption violation.
Interaction effect: Music and language conditions (Hypothesis 1)
The MANOVA results (see Table 1) showed a significant main effect of the music condition on the tasks, F(2, 159) = 5.03, p = .008; Pillai’s Trace = .059; partial eta squared = .059. No significant effect was observed for language or in the interaction effect between the music and language conditions.
Summary of MANOVA analysis: Breakdown of independent variables of music and language.
Effect of music familiarity on the various tasks (Hypothesis 2)
The tasks results were analysed post hoc using a series of two-way ANOVAs with the music conditions being the IV and each set of task results being the DV. Only the word memory task showed a statistically significant difference using ANOVA with a Bonferroni adjusted alpha level of .017, F(1, 160) = 9.94, p = .002, partial eta squared = .058.
Further analysis with a one-way ANOVA using Tukey HSD test showed that the word memory score was higher in the familiar music condition (M = 12.2, SD = 1.99) than in the unfamiliar condition (M = 10.8, SD = 2.02), p = .005, but no significant difference (p > .05) was observed when compared to the control “no-song” condition. The same trend was observed when using independent samples t-tests between familiar music (M = 12.2, SD = 1.99) and unfamiliar music (M = 10.8, SD = 2.92; t(110) = 3.15, p = .002). The magnitude of the differences in the means (mean difference = 1.38, 95% CI: .510 to 2.24) was moderate (eta squared = .07).
No significant effect of music condition was found on the arithmetic and reading comprehension tasks mean scores.
Analysis of the task mean scores (Table 2) showed highest scores on the arithmetic task (M = 13.2) and reading comprehension task (M = 5.67) in the no-song condition. Despite showing no significance difference when compared to the control no-song condition, the word memory task showed an interesting trend with its mean scores being the lowest in the unfamiliar song condition (M = 10.8) and highest in the familiar song condition (M = 12.2). Both the reading comprehension task and arithmetic tasks had the lowest scores in the unfamiliar song condition despite no significant difference with the other conditions (p > .05).
Analysis of the mean task scores for each song condition across all language conditions.
p < .05.
Effect of language on the tasks (Hypothesis 3)
There were neither trends nor a significant main effect of language in any tasks (p > .05).
Additional analysis of gender effect using standard multiple regression
Standard multiple regression analysis showed that only gender predicted reading comprehension scores, R2 = .07, F(3, 161) = 4.13, p = .007. Females performed better for the reading comprehension and memory tasks, while males performed better in the arithmetic task.
Discussion
In the current study, we sought to investigate the effect of the song language and song familiarity on three tasks (arithmetic, word memory, and also reading comprehensions). A between-subject design with randomized order of tasks was used for this study to reduce the effects of fatigue, habituation and training.
We found no significant effect of music familiarity and language when compared to the control “no-song” condition, thus we reject our first hypothesis that familiarity of music and language of lyrics will affect learning and task performance. Despite not finding statistically significant effects of background music on the tasks, we did observe an interesting trend of a negative impact of music on the whole, with the exception of familiar music.
For our second hypothesis on the effects of music familiarity, we found a significant difference for only the mean scores of the word memory task, hence we accept our hypothesis that familiarity of music affected word memory tasks but reject the hypothesis that music familiarity affected the arithmetic and reading comprehension tasks. We found an interesting trend that no music conditions had higher mean scores for both reading comprehension and word memory tasks performances (see Tables 1 and 2). Analysing the tasks separately, language and familiarity were not observed to have significant influence on the task scores with the exception of the word memory task where familiar music conditions had higher scores than unfamiliar music and quiet conditions.
On the third hypothesis that language of the song would have an effect, we did not find a significant effect; hence we reject our third hypothesis.
When considering that music loudness does not have an effect on task performance (Wolfe, 1983), our study adds to the literature that music has minimal impact on task performance. This is shown where we did not have significance differences (except for word memory task on familiar music to unfamiliar music) for our task scores but we observed interesting trends where no music conditions had better scores for two tasks in spite of the tasks’ difficulties being below undergraduate levels (the study population, therefore being less challenging). The background vocal music likely interfered with the phonological store, and thus affected the performance (Salamé & Baddeley, 1989). The arithmetic and reading tasks were cognitively demanding, and since the latter utilized the phonological store (Walczyk, 2000), the disruptive effects of vocal music were more pronounced (Baddeley, 1986) than word memory tasks. While the word memory task and music information are processed by the phonological loop, reading comprehension required more cognitive resources for identification, semantic access, inference generation and text elaboration (Walczyk, 2000). Thus our findings have trends that lend support that music interfered with cognitively demanding tasks (Baddeley, 1986; Etaugh & Michaels, 1975; Etaugh & Ptasnik, 1982; Fogelson, 1973; Mullikin & Henk, 1985; Salamé & Baddeley, 1989).
Although many studies have found that music impacts task performance negatively (Cassidy & MacDonald, 2007; Henderson, Crews, & Barlow, 1945) and positively, such as eliciting better reading comprehension (Hall, 1952) or arithmetic performance (Vaughn, 2000), we did not observe these at statistically significant levels. Such differences may be a result of the different populations measured confounded by possible arousal levels linked to increased cognitive performances (van den Bosch, Salimpoor, & Zatorre, 2013). As a control for our study population, we compared the performances of these tasks by gender. We found women to fare better on reading comprehension, supporting previous studies on higher reading self-efficacy (Horner & Shwery, 2002; Pajares, 2002; Raffaelli, Crockett, & Shen, 2005). Possibly enforced by female socialization practices that are typically shunned by boys, it is likely that females had better voluntary reading habits (Irwin, 2003). In the context of this study, the differences in performances by gender might have influenced their music preference, which does differ between genders (Hargreaves, Comber, & Colley 1995). Given that involvement and distractibility of the music were influenced by multiple factors – social and cultural upbringing, memories, and perceived meaning of music (Mitchell, MacDonald, & Knussen, 2008) – love songs such as “My Heart Will Go On” and “Volare” would have elicited different distractions, associations and arousal levels in the different genders. This may in turn affect attention (Huang & Shih, 2011).
Our study population is limited to psychology undergraduates. It is not known if students of all ages have the same susceptibility to the distractions of vocal music. In addition, there are subjective definitions of “familiarity” that can range from having heard the tune before to having memorized a song and being able to play it. This is a confounding factor as familiarity of vocal music can lead to different arousal levels (van den Bosch et al., 2013). At the same time, we acknowledge that, given the different thresholds of individuals, there is a possible influence of habituation to the music on our participants during the study. However, this was kept minimal given the random order of tasks and that the song length (~5-6 mins) is around the time given for a single task completion. As our study also did not separate learning and performance, it might be possible to do futher studies with music stopped while learning or when responding to the tasks.
Nonetheless, our findings were in agreement with previous studies in that the language of the music lyrics had little effect on learning and task performance (Colle & Welsh, 1976). Despite a trend of no music being better, there was no statistical grounding for us to conclude that music negatively interfered with task performance. These findings may be of interest to students who study in the presence of music.
Footnotes
Acknowledgements
We would like to thank James Cook University, Singapore, for the provision of the facilities for this work.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
