New technologies in the twenty-first century are increasingly affecting how people listen to music (Nill and Geipel 2010; Sloboda, Lamont, and Greasley 2009), and the availability and ubiquity of music today is greater than at any other time in history (North and Hargreaves 2008). Advancements in technology and digital information continue to accelerate, in all possible forms (Bontis 2011). For example, currently there are multiple ways in which music is delivered on the Internet. Legal means, as opposed to illegally distributed content, include à la carte (users pay a single fee for one song), tethered (users rent non-transferable songs for a limited period), album (popular, single fee for one album option), bundle (users can download a playlist, including multiple artists), streaming (increasingly popular, low cost, non-download option for consumers who want to explore a broad range of songs), and customised streaming (non-download service that allows users to compile their own programme of tracks based on genre/artist preference; for an overview, see Illing and Peitz 2006).

Although individuals may be relatively responsive to music in retail and commercial settings (see Yalcha and Spangenberg 2000; Areni and Kim 1993), developments in mobile devices (e.g., MP3 players and mobile smart phones) have led to better control of how, when, and where people listen to music (Heye and Lamont 2010; Juslin et al. 2008; O'Hara and Brown 2006; Hargreaves and Hargreaves 2004). The methods by which individuals are able to store, access, and obtain music have matched the growing popularity of media files such as MP3 (Kibby 2009), and it is important that research into consumption patterns and preferences keeps pace with and takes account of these changes (Gaunt and Hallam 2008).

Rapidly growing databases of music consumption, from live-stream services like Nokia Music,, or Spotify, or Nokia’s "Comes With Music" download service, are providing researchers with unprecedented insights into people's listening habits. The popularity of artists and tracks, and the factors that may contribute to musical fame, are of particular interest for purely academic reasons as well as commercial purposes. For example, Pachet and Roy examined whether the popularity of songs can be predicted based on extracted audio features and manual labels (2008). Their finding, that automated machine-learning techniques are as yet unable to accurately predict song popularity, contradicts earlier research suggesting that this is indeed possible (Dhanaraj and Logan 2005). Koenigstein and Shavitt presented an approach to predict music charts from search queries within the peer-to-peer sharing network Gnutella (2009). They showed that a song's peer-to-peer popularity highly correlates with ranking in Billboard charts, although whether this can be used in a predictive way is less clear. Motivated by a desire to predict the popularity of artists in specific regions of the world, Schedl et al. used a variety of approaches, such as page counts from Web search engines, user posts on Twitter, shared folders on Gnutella, and play-count data from (2010). Each approach overlapped only to a relatively modest extent, indicating artist-popularity is difficult to gauge and that a heuristic approach involving a combination of sources should perhaps be employed. Similarly, Schedl (2011) and Hauger and Schedl (2012) have explored the extent to which microblogs (e.g. Twitter) are used to communicate music-listening behaviour. They found that microblogging did indeed represent an important communication channel for revealing music-listening activities and the popularity of music artists in different regions at different periods of time, although the extent to which microblogs were used varied considerably from country to country, their accuracy as a predictor varied considerably.

Whilst being motivated by questions related to those outlined above, the aims of our study are distinct in a number of ways. Rather than determining the à priori popularity of songs from specific features within tracks, as Pachet and Roy (2008) have attempted to do, we are primarily concerned with investigating how the downloading "trajectories" of various genres differ, and the extent to which these change between countries. More specifically, we sought to identify which genres have markedly different consumption patterns, and in what ways the reception of artists differ throughout the world. In order to study this, four possible phases of a song's life are assumed to exist: growth (the initial rise of a song's popularity); maintenance (the extent to which a song is consistently downloaded); decline (the rate at which a song's popularity wanes); rebirth (specific musical and real-world events that might generate renewed interest in a song).

Events within the real world influence all four phases of songs' lives and the extent to which they are downloaded. Influences can be categorised as either musical or extra-musical in nature depending upon whether they relate specifically to the material and/or aesthetic aspects of tracks, or to the broader context in which they are consumed. For example, an artist winning a Grammy award or the release of a movie soundtrack or new album can be considered "musical," whereas a social event such as the death of an artist or a sporting event can be thought of as "extra-musical." As we will see, both musical and extra-musical events can have profound effects upon music consumption.

Underpinning this research is a five year data-sharing and cooperation agreement, begun in summer 2012, between McMaster University and the Nokia Corporation. Based on the compelling argument that data generated by the public should be returned to and reside within the public domain, the aim of the agreement was to establish a lab dedicated to the analysis of Nokia's global music download data from a sociological, cultural, and musicological perspective. Despite competitive market conditions, Nokia remains one of the world's largest manufacturers of mobile communications devices, with over 132,000 employees in 120 countries, and sales in more than 150 countries. Nokia's online music stores, some of whose data are analysed in this paper, allow users to download music directly onto a mobile device or computer. More specifically, purchasers of Nokia's smart mobile phones have gratis access to unlimited amounts of downloadable music, allowing users to explore their musical tastes unfettered by constraints of cost. Currently, some forty countries have Nokia music stores, representing all areas of the globe, and in total over 20 million tracks covering every conceivable music genre can be downloaded or live-streamed. The music-download data reported in this study relate to twenty-seven online stores, representing many geographically, socially, and economically disparate parts of the globe. Moreover, the current agreement between McMaster and Nokia provides access to data from 2007 to 2012, allowing time-dependent, longitudinal studies to be carried out. The database, basic querying method, and methodology are now briefly described.

Database and methodology

Currently, the database consists of over 180 million downloads, each containing various metadata relating to either the user or the track (this represents only a portion of Nokia's total database, and is therefore not indicative of market share). For example, user-metadata attributes include date, (local) time, (anonymised) user ID, user start date, user download count, country; track attributes include artist, genre, track title, duration, album, label, rightsholder, and so on. These data are arranged into a relational database management system (RDBMS) and queried using the open-source MySQL implementation of SQL (Weinberg, Groff, and Oppel 2010). The findings reported below are the result of initial forays into the database. Given the size of the database and the exploratory nature of our study, the research is to some extent selective and case-based rather than fully comprehensive and exhaustive.

Following the formulation of research questions appropriate to the database, a three-part heuristic approach was adopted that involved (1) information extraction using MySQL, (2) statistical analysis of extracted data, and (3) graphing and interpretation of results. Given a target such as artist, song, and/or date event, it is possible to pinpoint music downloads that relate to each category. For example, queries can be run that relate to the amount of downloads per month, week, day, hour, or even minute for a particular band. Similarly, information can be ascertained for specific users or subgroups of users, based, for example, on individual countries or download quantities. With respect to data analysis, this involved normalisation procedures such as calculating straightforward percentages, dispersion statistics for download distributions across a given time period, and correlating extracted data with either musical or extra-musical events. Lastly (and crucially), the findings were graphed to allow results to be displayed and interpreted "holistically." That is, given that many of the graphs represent large data volumes (in some cases many millions of downloads) over long time periods (months and years), a simple graphing method was adopted aimed at displaying the “big picture” that would be relatively easy to grasp.

Questions and results

Do songs belonging to different genres have different download trajectories?

Motivating this question was the notion that differences in the categorization of music based upon "style" may relate to differences in music-consumption behaviour. In other words, genre preferences—the result of perceiving musical difference—may reflect sociological phenomena relating to people's listening habits. Or put another way, if someone has a preference for a particular musical style, are they predisposed to acquiring songs in a different way from someone with different musical tastes?

To explore this question, three high-volume case studies relating to three musical genres were examined: a pop song, "Only Girl (in the World)" by Rihanna (2010); a rap song, "Love the Way You Lie" by Eminem (2010); and a rock song, "Sex on Fire" by Kings of Leon (2008). "Only Girl (in the World)" charted at number one in the United States' Billboard Hot 100 chart. "Love the Way You Lie" is Eminem's best-selling single and was number one on twenty-six record charts, including the United States' Billboard Hot 100 for seven weeks. And "Sex on Fire" gave Kings of Leon their first number one single in Australia, Finland, Ireland, and the United Kingdom, charting at the top spot on digital downloads in the UK, before its physical release. Each song, therefore, had considerable mainstream appeal within each genre.

Figure 1 shows download volumes as a percentage per month per track for a twenty-six month period following their releases; each track's total download volume is equal to 100 percent over the twenty-six month period shown in the figure. For the reasons referred to above, we were keen to observe the degree to which each song's pattern of growth, staying power, and decline differed. The pop track had a sixteen percent download-volume peak in month four, followed by a precipitous decline to six percent in month seven. Despite resurging around month eight, after twelve months download volumes had gradually fallen to about 2.5 percent. "Only Girl (in the World)" has yet to undergo rebirth, which, depending on the presence of a rebirth trigger, may never occur. The growth trajectory of the rap song was similar to the pop, steep rise following release. However, unlike the pop song, the rap song had a strong three month maintenance phase (months five to seven) before very rapidly declining in popularity. After twelve months download volumes had declined to less than two percent. In contrast to both the pop and rap song, the growth phase of the rock song was remarkably pedestrian, peaking only fifteen months after its release with just over seven percent of downloads. The benefit of a relatively slow growth phase, however, is that a song can remain culturally relevant for a longer period. At twenty-one months almost five percent of total downloads for the rock song were recorded, in sharp contrast to the pop and rap tracks for which less than one percent of downloads were recorded at this time.

Figure 1: Download trajectories for songs belonging to rock, pop and rap genres.

Download trajectories for songs belonging to rock, pop and rap genres.

The different download distributions between the genres as represented by these tracks are reflected in their respective standard deviations, which are: pop, 4.45; rap, 5.04; rock, 1.82. That is to say, the dispersion of the rock track is far more evenly spread throughout the twenty-six months than the pop and rock tracks. The rap track, despite its three month maintenance phase, has a higher standard deviation that the pop track due to its very rapid decline from month seven (15.1 percent) to month eleven (2.8 percent).

The marked differences between the tracks, as well as some similarities, may of course be due to a number of factors. With respect to similarities, the relatively high standard deviations for the pop and rap tracks, reflecting their steep rise and fall in popularity, could be the result to some degree of top-down processes relating to the music industry itself. For example, the extent to which local and national radio stations "push" a track will influence the degree and steepness of the growth phase; social media may play a role in spreading information about new releases by particular artists, influencing the speed of a song's growth phase. Common to many rap songs is the incorporation of a sung, melody-based chorus to contrast the rapped verses. In "Love the Way You Lie," by Eminem, the sung chorus features Rihanna as a guest artist, giving the song an intermittent pop-like quality. Arguably, therefore, the download trajectories of the pop and rap songs in the study may be linked because the rap song transgresses the genre boundary into pop (whether this is for commercial or aesthetic reasons is open to debate).

The points raised above, particularly in respect of top-down influences, seem less convincing, however, when viewed in the context of the rock song "Sex on Fire." As with the pop and rap songs, this Kings of Leon track was hugely successful and had highly visible mainstream media coverage in many countries, supported by a slick, visually alluring music video. Yet, despite this, its reception strongly differs in a number of key aspects. Most notable is the relatively flat trajectory over the twenty-six months shown in Figure 1—fourteen months had download volumes at or above four percent, pop had nine months above four percent, and rap had only seven months above four percent. Why, then, is the reception of a song from the rock genre so different? Assuming that the promotional mechanisms of the music and media industries were applied in similar ways for these songs—and the assumption, without performing an exhaustive search, is that this was largely the case—the difference in the reception of the tracks may be for either of the following reasons. First, consumers of rock music are a separate population with different music-behavioural traits (i.e. they only take to new songs relatively gradually). Second, consumers of rock music are not a separate population, but people in general are far more circumspect with respect to rock music than pop and rap. Both proposals are plausible; considerable further research will be required in order to cast light on this issue.

Possible differences with respect to musical styles can be further explored by asking whether there is a preference for older or newer tracks within each genre. If different genres are associated with different download behaviours, perhaps different genres will also be associated with differently aged tracks, for example older tracks with the rock genre than pop genre. To answer this, we calculated the mean and standard deviations of the original release dates of the ten most popular tracks in the database from the pop, rap, and rock genres. The mean release year for pop's top-ten tracks is 2009.5; rap, 2009.4; and rock, 2001.2. On the face of it, therefore, based on the mean year of release, substantially older tracks are preferred within the rock genre than the pop and rap genres. The dispersion of the preferred year of release also indicates possible important differences: for pop and rap the standard deviations are both 1.35, whereas for rock it is 11.37. In other words, the spread of release dates within top-ten rock tracks is large in comparison to pop and rap, underscoring the fact that rock is not simply a different musical style, but that the choices people make within it are distinct also. Figure 2 shows the mean release date per genre; error bars are +/- one standard deviation from the mean.

Figure 2: Mean preferred year of release for the top-ten rock, rap and pop tracks.

Mean preferred year of release for the top-ten rock, rap and pop tracks.

Do different countries have different download trajectories?

To further explore the cultural reception of songs and artists belonging to different genres, the question "do different counties have different download trajectories?" sought to ascertain whether a song with mass appeal, released in multiple countries at the same time, produces significantly different patterns of downloading. The song chosen for this study was "Bad Romance" by Lady Gaga, released in multiple countries onto Nokia mobile devices at the end of October 2009 (coinciding with the track's general release). Two Latin countries and two European countries were chosen for this study: Brazil and Mexico, and Germany and Italy.

Cultural homogeneity has long been a feature of discourses relating to globalisation. Views range from the belief that globalisation brings with it genuine integration (homogeneity), to an awareness and accentuation of cultural differences between peoples (heterogeneity). For example, writing in 1997, James H. Mittelman states, "globalization is a coalescence of varied transnational processes and domestic structures, allowing the economy, politics, culture, and ideology of one country to penetrate another"  (qtd. in Iriye 2002, 196). In contrast, sociologist Mike Featherstone wrote in 1995, "[o]ne paradoxical consequence of the process of not to produce homogeneity but to familiarize us with greater diversity, the extensive range of local cultures" (qtd. in Iriye 2002, 193). At least in respect of music, these opposing views can be investigated by observing the degree of similarity between download patterns. If globalisation has indeed led to the "coalescence of varied transnational processes and domestic structures" as Mittelman proposes, then the download trajectories of a song in different countries might be very similar. If, on the other hand, globalisation has so far led merely to an increased awareness of difference, download trajectories of a song might differ considerably from country to country.

Figure 3 shows download volumes for the song "Bad Romance" as a percentage per month for a three year period, beginning October 2009. Four countries are shown: Brazil, Mexico, Germany, and Italy; each country's total download volume is equal to 100 percent over the thirty-six month period shown in the figure.

Figure 3: Download trajectories for “Bad Romance” by Lady Gaga in Brazil, Germany, Italy, and Mexico.

Download trajectories for “Bad Romance” by Lady Gaga in Brazil, Germany, Italy, and Mexico.

To assess whether the four download distributions differed to a statistically significant degree, two-sample Kolmogorov–Smirnov tests (KS-tests) were performed between pairs of countries. The KS-test tries to determine if two datasets differ significantly, and has the advantage of making no assumption about the distribution of data (neither distribution needs to be normal, for example). More specifically, the test calculates the probability that two samples are drawn from the same parent distribution—the lower the p-value, the lower the probability of there being a parent association. Table 1 shows the p-values for all pairs of countries in the study (p-values for two-sample Kolmogorov–Smirnov tests performed on the download distribution of Brazil, Germany, Italy, and Mexico shown in Figure 3).

Brazil Mexico Germany Italy
Brazil / 0.32 0.15 0.06 [2]
Mexico 0.32 / 0.49 0.87
Germany 0.15 0.49 / 0.98
Italy 0.06 [3] 0.87 0.98 /

The lowest p-values (i.e. indicating least association between the distributions), are those for Brazil. This confirms what is clearly visible in Figure 3: the download trajectory of Brazil does not resemble that of the other countries in the study. However, the only pairing that differed to a (marginally) statistically significant degree is Brazil and Italy (p = 0.06). This could be due to significant differences in the cultural practices of music consumption in these two countries, indicating that the integrative effects of globalisation, as postulated by Mittelman, have yet to manifest themselves in this instance. Despite this finding, none of the p-values of the remaining pairings were significant, perhaps indicating that there are substantial crossovers in the cultural norms governing the acquisition of music and, by implication, perhaps other social behaviours too. This is most clearly suggested in the high p-value for Germany and Italy—two post-industrial democracies that have played a role in steering the European Economic Community, often referred to as the "Common Market," toward greater integration.

The above is of course not intended to prove or disprove the existence of shared cultural practices brought about through globalisation. Rather, the aim is to develop heuristic methods by which the complexities of cultures can be explored and examined. The approach adopted is therefore, by necessity, speculative, rather than based on a falsifiable hypothesis.

How do different events affect download trajectories?

The purpose of this question was to explore the effects upon downloading of two different categories of events, musical and extra-musical. As previously stated, events categorized as musical relate specifically to a musician’s artistic output, such as winning a Grammy award, whereas extra-musical events relate to non-artistic (social) incidents, such as the death of an artist. The musical event case reported is Adeles Grammy awards in February 2012. Extra-musical influences on downloading are studied with respect to the death of Michael Jackson in June 2009.


On 12 February 2012 Adele Laurie Blue Adkins crowned a meteoric rise to fame by winning six Grammys, including "Best Album of the Year" and "Best Pop Solo Performance." Awarded by the National Academy of Recording Arts and Sciences of the United States, Grammys are an annual celebration of outstanding achievement in the music industry. Moreover, given the considerable media attention the awards ceremony attracts, Grammys are also an important music-industry tool for promoting and marketing new and emerging artists. Of interest, therefore, is the effect the ceremony has on the download trajectories of award-winning artists, particularly Adele, who equaled the record for the most Grammy Awards won by a female artist in one night (Beyoncé Knowles won six Grammys in 2010). Does media attention in events like the Grammys translate into lasting changes in the download consumption of an artist's music?

To answer this, we investigated the download volume of Adele’s music prior to and following her success on 12 February 2012. Figure 4 shows download volumes for tracks by Adele as a percentage per day for a ten week period from 1 February to 18 April 2012. To model the effect of the Grammy Awards, the area bounded by the intersection of three lines, shown in Figure 4, was calculated: the vertical date line set to 12 February; the sloping linear trend line, modeling the drop in downloads following the peak on 14 February (R2 = 0.608); and the horizontal baseline, calculated from download volumes prior to 12 February. The point at which the linear trend line intersects the baseline indicates the date on which the effect of the Grammys on downloading ended, in this case 8 April, fifty-six days (eight weeks) after the ceremony took place. The area shaded dark grey, representing the positive effect of the Grammys, was then compared to the area under the baseline for the same period, shaded light grey. This revealed a 53% increase in downloads over the fifty-six day period from the 12 February to the date the download trend returned to the baseline on 8 April.

Figure 4: Download trajectory for Adele: 1 February—18 April, 2012. Dark shaded area shows increase in downloading following Grammy Awards, 12 February 2012.

Download trajectory for Adele: 1 February—18 April, 2012. Dark shaded area shows increase in downloading following Grammy Awards, 12 February 2012.

In addition to the analysis above, two features of the graph in Figure 4 are worth noting. First, there is a distinct rise in downloads immediately prior to the award ceremony, theninth to eleventh of February, probably due to anticipatory media speculation concerning Adele's possible success and related social-media "buzz." Second, the rise in downloading accelerated on the twelfth and thirteenth of February as news of Adele's success spread throughout mainstream media. The downloading peak of 2.75 percent on February fourteenth represents a three-fold increase from the baseline volume of 0.87 percent per day. These points are discussed in relation to Michael Jackson.

Michael Jackson

Following Michael Jackson's controversial death on 25 June 2009, and in the wake of mass media attention and speculation, the music industry worked hard to capitalise on the genuine outpouring of grief expressed across the globe. For example, on 26 and 28 October 2009 the documentary-concert album and movie This Is It was released, making its international debut in 110 territories. Consequently, Jackson became the best-selling albums artist of 2009, with some 8.2 million albums sold in the United States, and a total of 35 million albums worldwide (MSNBC 2010). In 2010, further interest in Jackson was generated with the release of the posthumous album Michael, a compilation of previously unreleased tracks. And in November 2011 the remix album Immortal, featuring The Jackson 5 and The Jacksons, was released after becoming the soundtrack to the Cirque du Soleil's Michael Jackson: The Immortal World Tour, which debuted in October 2011 in Montréal, Canada.

Jackson's death, and the subsequent release of Jackson-related music and films, allow both musical and extra-musical influences on downloading to be studied: "extra-musical" being his death, "musical" being the various activities of the entertainment industry with publishing and distribution rights to his work. Figure 5 shows download volumes for tracks by Michael Jackson as a percentage per month for a period of four years and three months from June 2008 to September 2012. Repeating the method applied to Adele, to model the overall effect on downloading of Jackson's death and subsequent industry output, the area bounded by the intersection of three lines, shown in Figure 5, was calculated: the vertical date line set to June 2009; the sloping linear trend line (R2 = 0.363); and the horizontal baseline, calculated from download volumes prior to June 2009. The intersection of the linear trend line and baseline, indicating the date at which baseline conditions were reestablished, is March 2012, thirty-three months after Jackson's death. When compared to the area under the baseline, shaded light gray, the dark shaded area, representing the positive effect upon downloading of Jackson’s death, reveals a 230 percent increase over the thirty-three months from June 2009 to the date the download trend returned to the baseline in March 2012.

Figure 5: From the Visual Model of Paul's Churchyard, the Cross Yard. Constructed by Joshua Stephens and rendered by Jordan Grey.

From the Visual Model of Paul's Churchyard, the Cross Yard. Constructed by Joshua Stephens and rendered by Jordan Grey.

From the graph in Figure 5, both musical and extra-musical effects on downloading are clearly apparent. First, and most striking, is the extraordinary peak in downloading in the weeks immediately following Jackson’s death: the peak of 13.43 percent in July 2009 represents almost a seventeen-fold increase from the baseline volume of 0.8 percent per month. Also conspicuous is the peak relating to a second extra-musical event, the first anniversary of Jackson's death in June 2010. Here, the 4.82 percent peak is a six-fold increase from the baseline. The three musical events referred to above, the release of This Is It, Michael, and Immortal, are also visible as peaks in downloading, 3.37 percent, 3.13 percent, and 2.01 percent respectively. Given the considerable efforts of the music and film industries in the months and years following Jackson's death to bring to market products that met the demands of his huge fan base (McCartney 2010), it is perhaps surprising that these downloading peaks are all lower than the extra-musical peaks relating to the artist's death. Simply put, with respect to downloading, Jackson's death appears to have motivated people to acquire his music to a far greater degree than, for example, the release of new material by the artist.

Through spontaneous actions, such as the creation of impromptu shrines or candlelight vigils, outpourings of grief over the death of popular public figures, such as Michael Jackson in 2009, powerfully demonstrate the ability of individuals within society to behave in a collective manner. Despite the role the media plays in such events, as especially noted by Kear and Steinberg (2002, 3-4) in relation to the death of Lady Diana Spencer, crucial to those participating in grieving are three main elements: the overwhelming empowerment of grief; the belief that the presence of the deceased can somehow be felt; and the understanding that the place of memorialisation is special. According to Clark and Franzmann, "the strength of grief, the power of presence and the importance of place allows ordinary people to assume and, therefore, challenge the authority of...the government as official purveyors and regulators of mourning ritual" (2006, 579). That is to say, through collective action, individuals assume the role of actors whose actions have both meaning and power.

The notion that at times of public grief individuals become proactive participants in an event rather than merely recipients of officially sanctioned mourning, may partly explain the finding, shown in Figure 5, that the extra-musical events impacted downloading more than musical ones. More specifically, the musical events—in this case, albums released through official industry channels—were "top-down" commodities that people consumed rather than contributed to, whereas the extra-musical events—Jacksons death and commemoration a year later—involved public grieving and remembrance, calling for active participation. By necessity, downloading music requires motivation and action. Arguably, given the active "grass-roots" participation of people in grieving, the extra-musical events associated with Jackson's death promoted both motivation and action with respect to downloading to a greater extent than the more passive musical events.

The relatively weak effect of musical versus extra-musical events on music downloading is substantiated in the analysis of Adele downloads following her triumph at the 2012 Grammy Awards. Despite Adele's unprecedented success, and the ensuing accolades bestowed upon her by the music industry and media, the overall effect lasted only eight weeks and increased downloads by a little over fifty percent. In comparison, the consequences of Jackson's death lasted almost three years (thirty-three months), and resulted in a 230 percent increase in downloads, supported by album and film releases. Admittedly, Michael Jackson and Adele are very different artists, and so the comparison is artificial. Jackson’s career spanned four decades, developing from boyhood stardom to remarkable adult success. Adele, on the other hand, is relatively early in her career, and, unlike Jackson, owes her success more to the Internet than a dynastic musical family (Adele gained her first recording contract after a friend posted her demo on the social-networking service Myspace in 2006. Michael Jackson's full professional career began at the age of eleven when the Jackson 5 signed to Motown in 1968; they issued their debut single "I want you back" in October 1969 (Huey 2014). However, despite these differences, the analyses above clearly suggest that the degree to which people feel invested and involved in an event, and the extent to which it promotes either passive or active engagement, strongly influences their music-downloading behaviour.


This paper has sought to investigate some of the complex issues surrounding the consumption of music through downloading. We began by outlining three phases within the life of a song: growth, maintenance, and decline. In addition, a fourth possible phase was identified, rebirth. Perhaps the clearest example of the fourth phase can be seen in Figure 5 above. Here, the trigger for the rebirth of many songs by Michael Jackson was, paradoxically, the death of the artist. The lasting legacy of this extra-musical event, with respect to the consumption of his music, may well take many years to reveal itself. However, given the veneration from fans and the academic interest he continues to receive (Hidalgo and Weiner 2010), it is likely that the rebirth phase of much of his music will last for many years.

The finding that the rock song, "Sex on Fire" by Kings of Leon, had a different download trajectory to the pop and rap-genre tracks, and that older tracks are preferred within the rock genre, raises the possibility that rock music is either listened to by a subset of people with different listening behaviours or that rock music is listened to in a different way by a majority of people. Over the past decade a number of studies have sought to link musical preference with personality (for overviews see North and Hargreaves 2008; Clark, Dibben, and Pitts 2010). While these studies are not entirely consistent, sufficiently consistent trends have emerged to suggest that it is possible, in theory at least, to deduce something about character based on musical preferences. Future research might therefore seek to explore whether it is possible to deduce something about the national "character" of a country with respect to attributes such as self-esteem, creativity, attitude-to-work, outgoingness, gentleness, and how "at ease" people are within a country.

The interaction between song and country was investigated in relation to Lady Gaga's international hit "Bad Romance." KS-tests confirmed that Brazil had a substantially different download trajectory to the other countries in the analysis—Mexico, Germany, and Italy. While we speculate that this may reflect the existence of distinct cultural practices in Brazil, which have yet to be ameliorated through the assimilating forces of globalisation, additional local factors may be responsible for the observed differences. For example, one possibility is that the dominance of pop is challenged within Brazil, that the musical "space" is crowded thereby pushing out artists who hold centre stage elsewhere. Brazil certainly is highly diverse musically, possessing a rich heritage created from a complex mix of different cultures; the ability of artists to penetrate its musical culture may therefore be severely curtailed. However, whether a crowded musical space is sufficient in and of itself to produce the radically different trajectory observed in Figure 2 is questionable. Consequently, the issue of Lady Gaga and the reception of pop music in general within Brazil remains open.

Finally, musical and extra-musical influences were investigated in respect of Adele's triumph at the 2012 Grammy Awards and the death of Michael Jackson. A comparison between the two case studies, in terms of degree and longevity of effect, revealed the potency of grief (however vicariously felt) as a motivator for action, in this case the action to seek and download tracks by a specific artist. A lesson that can be learned from this is that when people feel they are participating in something with collective meaning, then the top-down delivery of products from the music industry to the public is likely to be more successful. That is to say, that the marketing of artists and their music gains greater traction in situations where a passive-reception model is replaced with an active-involvement model. Participatory fan clubs are just one way in which bands attempt to achieve this, and in this respect, tours and live performances—the mainstay of popular music culture in the past—will undoubtedly continue to be important.

Our research is based on a music download database provided by Nokia, covering a five year period from 2007 to 2012. The studies presented in this paper are a selection of our initial forays into the data, and much work remains to be done before a fuller picture of music consumption can emerge. From the perspective of the digital humanities, these data clearly contain a wealth of information that will continue to require the application of innovative approaches and statistical methods to tease out the stories they contain. The lab in which the research is done is fortunate to be supported by funding from the Social Sciences and Humanities Research Council of Canada, which has allowed us to foster international and domestic collaborations, and to train students in relational database management systems. It is one thing to study significant patterns of music listening, and quite another to understand how they may relate to broader social issues and developments. With the expertise of collaborators from interdisciplinary backgrounds and the input of well-trained students, we anticipate making further progress not only with respect to the dynamics of music consumption, but also its meaning and significance at a human, cultural level.


The authors would like to express their thanks to the following people and organisations, who in various ways have contributed to the research presented in this paper: Nokia Music (Bristol, UK); Dr. Nick Collins (University of Durham, UK); Jotthi Bansal (Research Assistant, Digital Music Lab in Association with Nokia). This research was supported with an Insight Development Grant from the Social Sciences and Humanities Research Council, awarded to the first author. Findings referred to in the paper were first presented at the 2013 annual conference of the Canadian Society for Digital Humanities / Société canadienne des humanités numériques, at Victoria, British Colombia.


[1] "Every Track You Take" is a reference to "Every Breath You Take," a song by The Police on the band's 1983 album Synchronicity, written by Sting.

[2] Marginally significant

[3] Marginally significant

Works Cited

Areni, Charles S., and David Kim. 1993. "The influence of background music on shopping behavior: classical versus top-forty music in a wine store." Advances in consumer research 20(1): 336-340.

Bontis, Nick. 2011. Information Bombardment: Rising Above the Digital Onslaught. Institute for Intellectual Cap, Hamilton.

Clarke, Eric, Nicola Dibben, and Stephanie Pitts. 2010. Music and Mind in Everyday Life. Oxford: Oxford UP.

Clark, Jennifer, and Majella Franzmann. 2006. "Authority from Grief, Presence and Place in the Making of Roadside Memorials." Death Studies 30(6): 579-599.

Dhanaraj, Ruth, and Beth Logan. 2005. "Automatic Prediction of Hit Songs." In ISMIR: 6th International Conference on Music Information Retrieval, London, UK, September 11-15, 488-491.

Gaunt, Helena, and Susan Hallam. 2008. "Individuality in the learning of musical skills." The Oxford Handbook of Music Psychology, edited by Ian Cross, Susan Hallam, and Michael Thaut. 274-284. Oxford: Oxford UP.

Hauger, David, and Markus Schedl. 2012. "Exploring Geospatial Music Listening Patterns in Microblog Data." In AMR 2012: 10th international workshop on Adaptive Multimedia Retrieval, Copenhagen, Denmark, October 24-25, 1-15.

Heye, Andreas, and Alexandra Lamont. 2010. "Mobile Listening Situations in Everyday Life: The Use of MP3 Players while Travelling." Musicae Scientiae 14(1): 95-120.

Hidalgo, Susan, and Robert G. Weiner. 2010. Wanna Be Startin'Somethin': MJ in the Scholarly Literature: A Selected Bibliographic Guide.The Journal of Pan African Studies 3(3): 14-28.

Huey, Steve. 2014. "Artist biography of Michael Jackson." All Music. Retrieved May 30, 2014.

Illing, Gerhard, and Martin Peitz, eds. 2006. Industrial Organization and the Digital Economy. Cambridge, Mass: MIT Press.

Iriye, Akira. 2002. Global Community: The Role of International Organizations in the Making of the Contemporary World. Berkeley, CA: University of California Press.

Juslin, Patrik N., Simon Liljeström, Daniel Västfjäll, Gonçalo Barradas, and Ana Silva. 2008. "An Experience Sampling Study of Emotional Reactions to Music: Listener, Music, and Situation." Emotion 8(5): 668.

Kear, Adrian, and Deborah Lynn Steinberg, eds. 2002. Mourning Diana: Nation, culture and the performance of grief. New York, NY: Routledge.

Kibby, Marjorie. 2009. "Collect Yourself: Negotiating Personal Music Archives." Information, Communication & Society 12 (3): 428-443.

Koenigstein, Noam, and Yuval Shavitt. 2009. "Song Ranking based on Piracy in Peer-to-Peer Networks." In ISMIR: 10th International Society for Music Information Retrieval Conference, Kobe, Japan, October 26-30, 633-638.

McCartney, Anthony. 2010. "A Year After Michael Jackson's Death, Fan Base Remains Committed to Keeping Memory Alive." Associated Press, June 23. Accessed September 20, 2013.

MSNBC. 2010. "Jackson Sells 35 Million Albums Since Death – Entertainment," Access Hollywood, June 25. Retrieved 20 September 2013.

Nill, Alexander, and Andreas Geipel. 2010. "Sharing and Owning of Musical Works Copyright Protection from a Societal Perspective." Journal of Macromarketing 30 (1): 33-49.

North, Adrian C., and David John Hargreaves. 2008. The Social and Applied Psychology of Music. Oxford: Oxford University Press.

North, Adrian C., David J. Hargreaves, and Jon J. Hargreaves. 2004. "Uses of Music in Everyday Life." Music Perception 22(1): 41-77.

O'Hara, Kenton, and Barry Brown, eds. 2006. Consuming Music Together: Social and Collaborative Aspects of Music Consumption Technologies 35 (Spring): 7-22.

Pachet, François, and Pierre Roy. 2008. "Hit Song Science Is Not Yet a Science." In ISMIR: Ninth International Conference on Music Information Retrieval, Drexel University, Philadelphia, Pennsylvania, USA, September 14-18, edited by Juan Pablo Bello, Elaine Chew, and Douglas Turnbull, 355-360.

Schedl, Markus, Tim Pohle, Noam Koenigstein, and Peter Knees. 2010. "What's Hot? Estimating Country-specific Artist Popularity." In ISMIR: 11th International Society for Music Information Retrieval Conference, Utrecht, Netherlands, August 9-13, edited by J. Stephen Downie and Remco C. Veltkamp, 117-122.

Schedl, Markus. 2011. "Analyzing the Potential of Microblogs for Spatio-Temporal Popularity Estimation of Music Artists." In IJCAI: International Workshop on Social Web Mining, Barcelona, Spain.

Sloboda, John A., Alexandra Lamont, and Alinka Greasley. 2009."Choosing to Hear Music: Motivation, Process and Effect." The Oxford Handbook of Music Psychology, edited by Ian Cross, Susan Hallam, and Michael Thaut. 431-440. Oxford: Oxford UP.

Weinberg, Paul N., James R. Groff, and Andrew J. Oppel. 2010. SQL: The Complete Reference. 3rd ed. New York, NY: McGraw-Hill.

Yalcha, Richard. F., and Eric Spangenberg. 2000. "The Effects of Music in a Retail Setting on Real and Perceived Shopping Times." Journal of Business Research 49 (2): 139–147.

Valid XHTML 1.0!