A university for the real world

Football fandom in the Brazilian Twitterland

Posted In News, Projects - By On Tuesday, August 20th, 2013 With 0 Comments

Most people around the world know how passionate Brazilians are about football. More precisely, around 80% of the population there is at some level a football fan, a recent survey suggests (Pluri, 2013). In the Southeast, the region that concentrates most professional clubs playing in the top division, only 16% declared no support for any particular club. Also, Brazilians are rather active on major social media platforms. A recent post by my CCI colleague Darryl Woodford indicates the location of a recent 1 million Twitter users. Brazil is well represented, even though Twitter is actually only the 4th most accessed social media platform by Brazilians (Hitwise, 2013). Facebook, YouTube, and the Q&A website Ask.fm, respectively, are the top three. Orkut, once the most popular social networking platform there, is now the 5th.

So, Twitter is definitely not the most popular social media site there, but it has been central for football fandom communities since 2009, when Twitter started to grow in Brazil. The interest-based nature of Twitter conversations and the fact that key fans in the clubs’ communities migrated from Orkut to Twitter both may have contributed to this. Furthermore, most sport journalists have accounts on Twitter, making it more attractive for devoted fans to engage in “expert” conversations.

Altogether, this makes the Brazilian Twitterland an interesting space to understand what football fans have been up to online. As a departing point, Twitter data is able to show a good snapshot of the online everyday practices of those fans. And this is actually a rather non-explored research topic. Even though Twitter, Facebook, YouTube, and other spaces have been used for sport-related conversation and sport fan production, those practices have not received substantial attention from either social sciences or humanities scholars so far (Gibbons & Dixon, 2010). Different from the pop culture fandom area, where the implications of the Internet and other ICTs have been widely explored in the last decade, there is not much empirical work around the formation of online sport fandom communities.

Particularly about Twitter, two papers approaching football have recently come to my attention, showing that this topic is starting to spark interest. The first one, titled “Twitter and Sports: Football Fandom in Emerging and Established markets”, is a collaboration by CCI colleagues Axel Bruns and Stephen Harrington with the German-based researcher Katrin Weller. This paper will be published soon in the collection Twitter and Society and approaches football clubs in Australia, Germany, and England. The second one, “Changing the Game? The Impact of Twitter on Relationships between Football Clubs, Supporters and the Sports Media“, by John Price, Neil Farrington, and Lee Hall, was published very recently in a 2013 edition of the journal Soccer & Society. Both papers shed light to how professional clubs are managing their communication with fan bases on Twitter. And, importantly, both studies are based on empirical data from Twitter and other sources.

My PhD project and this brief post intend to look at sport-related conversations on Twitter from another angle. I am focusing on fan conversations and fandom activities performed online. So, instead of collecting data posted to or from the official accounts of clubs, I am doing it by keywords, particularly, clubs’ names and nicknames.

It is a challenge to gather such an archive. Some clubs in Brazil have names that are also common nouns. Furthermore, to find out which nicknames and other expressions fans use to talk about their clubs is a challenge in itself. Not to mention the fact that some names/nicknames are also words in other languages than Portuguese. After a couple of tests, I came up with this method that seems to be working more or less fine:

1) I use the name and one nickname for each club; the nickname was chosen after I tested some options and the one that was able to capture the most/best data was adopted;

2) I am applying a series of filters after collecting the data, starting from considering only messages in Portuguese;

3) Another important filter is a list of words to get only football-related tweets (for instance, Flamengo, a famous Brazilian club, is also the name of a suburb in Rio). This list includes the name of current players, coach and other important staff of each club when processing the data of that particular club, plus a list of usual words related to football such as goal, championship, referee and so on. Everything that does not have any of these expressions is excluded at this point;

4) And last but not least, a filter to exclude football-related tweets that are however not related to a particular club. For instance, two professional clubs in Brazil have the same name: one is Atlético-MG and the other one is Atlético-PR, with the acronym at the very end indicating the state from where they are from. So, the Atlético-MG’s data had to be filtered to not include “paraná” or “paranaense”, expressions used to refer to the other club. And, in this particular case, the use, for instance, of “Atlético-MG” as a keyword is not a good option because supporters of this club do not call it “Atlético-MG”.

After testing a few times the methods for collecting/filtering tweets, I ended up with a dataset comprised almost entirely of football-related tweets and a dataset that also seems to match the size of the fan bases of each club. My analysis here includes the 12 football clubs with the largest fan bases in Brazil, and the data was collected using yourTwapperkeeper.

Initial data and brief considerations

Below, I present some charts with data from the first week of the top division of the 2013 Brazilian Championship (the série A of the Campeonato Brasileiro or Brasileirão). Actually, it is a six-day period, containing 516,444 tweets after all the filtering.

 

 

The first chart is basically the amount of messages by day. This particular chart is deeply influenced by the transfer of Neymar to Barcelona. Neymar is a young talented Brazilian footballer who announced he was moving from Santos to Barcelona exactly in this week. This is the reason why there are so many messages about Santos, which is a popular club but has only the 7th largest fan base in Brazil. The other clubs’ conversations just follow the pattern: when the team is on the pitch, fans post more messages.

The second chart (below) is more interesting because it reveals a clear pattern in the fan conversations: a decrease in the proportion of URLs being shared on match-days. It occurs with all clubs every time they played in this interval (most of them played twice).

 

 

Thus, this particular chart demonstrates the nature of the conversations when the team is playing: more spontaneous, with a celebratory or protest vein. Also, when considering the type of message they are sharing in each particular moment (the chart below is an example using data for Flamengo), the proportion of original tweets tends to be higher on match-days, and the level of genuine replies tends to decrease, on the other hand. However, this particular pattern is not a rule, because there seem to be some differences depending on the result of the match and whether the team is playing at home or away.

 

 

In any case, on match-days there is a propensity to less interaction. Furthermore, my method is not able to capture messages that do not use the name or nickname of the club. And it is important to note that fans very often post messages celebrating goals and complaining about players with no mention to the club’s name or nickname. It is likely that this type of message increases on match-days.

Another series of charts I created showcases domains shared by users per club. The image below is the general chart (for an enhanced view, click on the image), with all domains shared in all conversations. The chart for each club can be accessed by clicking on the image. Particularly, the charts with club-specific data show, for instance, those blogs and other user-generated content spaces that have more significance for each community. The presence of traditional media outlets is very strong in those charts, revealing continuity patterns concerning sports audiences in a new media environment.

 

I am also trying to visualise which domains are most shared on a daily basis. The chart below is an example of what I am trying to do (again, click on the image for a better visualisation).

So, for instance, in this chart, Flamengo’s fans are sharing Twitter (mostly, photos), Instagram, and Facebook links in the match-day – Flamengo only played on the 26th in this period. Furthermore, the presence of Globoesporte.globo.com, the online portal of the largest media conglomerate in Brazil, slightly decreases in the day preceding the match. Now, I am comparing distinct clubs over a longer period to understand when they are sharing links to YouTube, Instagram, Foursquare and other platforms. But some interesting phenomena have already emerged: for example, blog posts seem to be shared mostly in the days preceding the matches and on the day immediately after, a pattern that follows the routines of most fan blogs. Instagram and Foursquare links are mostly shared on match-days, and YouTube videos seem to be shared in a higher proportion on the day immediately after matches.

Wrapping up

The most important aspect that I got from this preliminary analysis is that some fan initiatives are rather influential over the fan bases they talk about. Less than marginal, sometimes their significance over the conversations is suggestive. For instance, take the case of Atlético-MG as an example. Atlético-MG’s official account (@sitedogalo) has around 100,000 followers. Fan initiatives such as @cam1sado2e (a type of collective site, where fans publish videos, photos, short stories, anecdotes, chronicles, biographical stories and so on), and @webradiogalo (a fan produced radio/TV that does live commentary during the matches) have almost 15,000 followers each, a considerable amount – and they are probably not buying followers. Yet, @webradiogalo’s YouTube channel has as many views as some clubs’ official accounts – its videos were watched around 2.2 million times, which is similar to how many times Fluminense’s official videos were played (2.7 million), for instance. And those two initiatives are not unique: all fan bases have things like those being produced and largely shared.

Now, I am trying to analyse the networks themselves and explore hierarchy issues in such communities. Particularly, those key fans I mentioned above, who have/had important positions in Orkut communities (as managers or creators), seem to be the most influential fans on Twitter too. This particular aspect highlights how some of those fans have been involved with those communities for a long time, a few since 2004, the year when the most popular football fan communities on Orkut were created in Brazil. Those are the fans I am planning to interview for the qualitative part of my research.

 

Technical note: most data for this post was processed using scripts developed at the CCI. Particularly: the first, second and third charts (metrify.awk by day); forth chart (urlextract.awk, urlresolve.awk and urltruncate.awk); and filtering (filter.awk and filterinverse.awk). More at Mapping Online Publics (http://mappingonlinepublics.net/category/processing/).

 

About the Author

- Ana Vimieiro is a PhD Candidate at the ARC Centre of Excellence for Creative Industries and Innovation at the Queensland University of Technology (QUT). She is investigating sport fandom activities on Twitter in her doctoral project, and she is particularly interested in participatory practices that take place in online football communities in Brazil. Ana’s research interests include sport fandom, fan activism, digital culture, participatory culture, social media, and methodological innovations.

Leave a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>