#grexit: Meme Diffusion Across Platform and Location

Team Members

Lisa Madlberger, Mylynn Felt, Ana Pop Stefanija, Ana Rantovic, Rishabh Dara, Emilio Fernández Peña, Luigi Di Martino

Introduction

Our intent is to examine representations of the Greek crisis on the internet. “The web is more social than informational” (Rogers, 2013, pp. 155-156). Rather than focusing on news channels and mainstream media representations of this issue, we focus on the spread of memes on various social media platforms.

Memes

  • Three main attributes of memes (Shifman, 2013, pp. 364-365)

    • “cultural information that passes along from person to person, yet gradually scales into a shared social phenomenon” (pp. 364-365)

    • “reproduce by various means of imitation” (p. 365)

    • “diffusion through competition and selection” (p. 365)

This definition of memes does not tie the concept to images. Other forms of cultural information can pass from person to person, gradually scaling into a shared social phenomenon while being reproduced and diffused. While our data set included many images, we focused on hashtags to trace the reproduction and diffusion of Greek crisis memes.

  • “I suggest looking at Internet memes not as single ideas or formulas that propagated well, but as groups of content items that were created with awareness of each other and share common characteristics. Going back to Dawkins’ original idea—that memes are units of imitation—I find it useful to isolate three dimensions of cultural items that people can potentially imitate: content, form, and stance.” (Shifman, 2013, p. 367)

  • With Shifman’s definition of memes in mind, we followed what Knobel and Lankshear did by creating an initial “meme pool” using “different kinds of well-known online search engines” with the seed meme of #grexit as our initial search term (Knobel & Lankshear, 2007, p. 202).

Cultural Context

  • #grexit = potential Greek exit from Eurozone due to excessive debt

  • Data grab occurs between bank closures and referendum (30 June to 2 July 2015)

    • Data not confined to this time frame

Diffusion of #*exit Meme

It is interesting to see that the seed meme grexit has been adapted for different purposes and after performing a co-hashtag analysis on Twitter we could identify over 140 different variations.

Twitter Co-Hashtag Analysis Filtered for *exit

Research Questions

The goal of our research was to answer the following questions

  1. How are different platforms inter-related?

  2. How can we study the relationship between platforms?

  3. How does the seed meme #grexit diffuse across platforms and between countries?

Methodology

We had the ambitious goal to examine five different platforms. The traditional way we searched for is to search for Hashtags. However, we realized that hashtag-search is limited when applied to YouTube and Facebook. That is why we explored other type linkages between platforms, which leads us to our primary research question.


RQ1: How are different platforms inter-related?

The first thing we looked at, was to what extend do platforms support cross-posting content to other platforms in their native interfaces. Instagram, for example, allows postings to Facebook, Twitter and Tumblr. After doing this mapping we could identify that there are different classes of platforms. The secondary platforms (Youtube, Instagram and Tumblr) feed the primary platforms (Facebook and Twitter) as well as each other, while the primary platforms mainly receive but do not share content.


Facebook

Twitter

Instagram

Youtube

Tumblr

Allows posting on Facebook

Yes

Yes (Sync)

Yes

Yes

Yes

Allows posting on Twitter

Yes for Pages; No for Profiles

Yes

Yes

Yes

Yes

Allows posting on Instagram

No

No

Yes

No

No

Allows posting on Youtube

No

No

No

--

No

Allows posting on Tumblr

No

No

Yes

Yes

Yes

RQ2: How can we study the relationship between platforms?

After identifying the different types of relations, the second question we asked ourselves was “How can we study the relationship between platforms?” Using the simple keyword “grexit” as a seed, we collected five datasets for each of the platforms. We identified three different techniques to study the relationships between the platforms:

  1. URLs

We extracted the URLs from each of the datasets and quantified the number of links pointing to the other platforms. In this graph, each node represents a platform. As shown in the figure below, the big blue curve indicates Instagram links found on Twitter.

The size of the nodes relates to the number of in-degrees. Nodes for Tumblr and Instagram are really small; that corresponds what we previously found, which means they primarily feeders. In contrast, Facebook and Twitter are much bigger nodes which indicates a lot of platforms are publishing their content to these platforms. The unexpected anomaly is the relationship between YouTube and Facebook. What we expected is that we would find a lot of links from YouTube on Facebook, but we were surprised about the number of links to Facebook on YouTube.

Gephi - Cross Platform.png

2) #Prashtags: Platform-Relevant Hashtags

The second thing we looked at were so called Platform-Relevant Hashtags that we coined as “Prashtags”. Prashtags are hashtags that carry information about another platform. You see examples here, and we found them for all the platforms. Nevertheless, the Instagram ones were most prevalent. Platforms using Prashtags include the following: Instagram, Twitter, and Tumblr. Platforms most referred to by Prashtags include the following: Instagram, Twitter, YouTube.


As Gerlitz and Rieder (2013) note, “The use practices of platform features on Twitter are, however, not solely produced by users themselves, but crystallise in relation to wider ecologies of platforms, users, other media, and third party services (Burgess and Bruns), allowing for sometimes unanticipated vectors of development.” They cite the development of the Twitter retweet button as an example of user practice influencing platform development. In the case of prashtags, user practice textually connects platforms by correlating them with referent hashtags.


https://lh5.googleusercontent.com/vSDWUrATsHlvz5GTifNPeLKl5n7RcwFVfMwM_qa35rGEHGUvU7Ux-VIL1iqRSLS_Gygqbrs9ihUJ5eXh_2vtcjRhfmthQrl4HVanHE4YuCSgJd6UDu2Lg2qr4k1XvQzqS-Qy_T8MNw

3) Source Platform

Some of the APIs allow you to retrieve the source platform of a post. We looked at the tweets in our dataset that originated from one of the other four platforms that we studied. We can see that relative to the other platforms, most tweets originated in Tumblr.

https://lh3.googleusercontent.com/1T_AE1YjeWdZdU0A_9fJ63OQoiXFTUmnYMOvSRG1zfdtEnQHWQLBvYQc9g0EfNp68Ni2WjSMvIHWLI--NDq2OyV106Whs8n1md7UTHX6u3EGI0ty5AisMxrQztwiTdQPeLDg1bj7TA

RQ3 How does the seed meme #grexit diffuse across platforms and between countries?

We conducted four different analyses. As you can see, some forms of analysis were limited to certain platforms.

Twitter

Instagram

Facebook

Youtube

Tumblr

Co-Hashtag

x

x

x

Location

x

x

Images

x

x

Frequency over Time

x

x

x

x

x


Co-Hashtag

We conducted a co-hashtag analysis for Twitter, Instagram and Tumblr. There’s a great degree of overlap of terms between the platforms; these are the 20 most common co-hastags appearing with #grexit on Tumblr, Instagram and Twitter. cross-Platform Co-Hashtag Triangulation:

Common Hashtags cross-platform.jpg

Location

Next, we used location. We made a list of nine different ways to derive locations from posts and evaluated them for each platform. Of course, using GPS tags is the most accurate but not all platforms support this.

Facebook

Twitter

Instagram

YouTube

Tumblr

Data includes a GPS-Tag

No

Yes

Yes

No

No

Location-relevant Hashtag

Yes

Yes

Yes

No

Yes

Locations mentioned in Text

Yes

Yes

Yes

Yes

Yes

User-profile Location field

No

Yes

No

No

No

User time-zone setting

No

Yes

No

No

No

Language field in posting

No

Yes

No

No

Language field in user-profile

No

Yes

No

No

No

Regional domain, e.g. google.at

No

No

No

No

No

IP-address information

No

No

No

No

No



We then generated heatmaps using google map’s API to identify areas where the issue was being discussed.#grexit in Europe

Twitter

Displaying TwitterGrexit.JPG

In the Twitter discussion we can see a hotspot in Britain, whereas the actual event is taking place in Greece.


Instagram

InstagramGrexit.JPG

ter

#NoGrexit in Europe

If we look at the oppositional hashtag, we can observe a lot of activity in Germany. While on Instagram the #NoGrexit activity is focused within Greece.

Twitter

TwitterNogrexit.JPG

Instagram

InstagramNogrexit.JPG


#Brexit in Europe

Not surprisingly, we see #Brexit has been trending in Britain. On Instagram we can see that Brexit is also trending in Denmark, which possibly highlights Denmark’s EU skepticism.

Twitter

TwitterBrexit.JPG

Instagram

InstagramBrexit.JPG
The heat maps highlight that platforms and national contexts are closely related to each other.

Images

Next we looked at Images. We could retrieve Images from Instagram and Tumblr. Here we plotted the images retrieved from Instagram chronologically. The gaps in between the groups indicate when Europe was sleeping. We can observe an increase of images, over the time.

Chronology of Instagram #grexit

ImagePlot - Instagram Time vs Hue SD.png

Chronology of Tumblr #grexit

On Tumblr we found fewer images tagged with #grexit, but we could observe an increase in the frequency.

ImagePlot - Tumblr Time vs Hue Median.png

Frequency over Time

In a last step, we looked at the number of posts, over time. This was the only analysis we could do across all of the platforms.

Earliest Retrieved Data per Platform

When retrieving data from different platforms, we realized that the data sets covers different time frames. For some we could go back to 2012. For Twitter we could just go back three days until we hit the limits. The following chart represents data was retrieved 1 July 2015.

https://lh4.googleusercontent.com/mrNVNA-oqsnH4TFeKTg1et2D1HBgrW0-v-2TyRlVCJYpqjNLwrBuvixx3dvLl6fnCJD7_a6n08xIvSKpxAsbTRuUPoQPBX5_B6n2nmNerO2s-uzhNgJFcxGpowBy8T4OEklGQOGqzA

If we focus on the last one month. We should be careful while interpreting this data because what looks like a peak might due to a tool hitting an API’s threshold. The following chart focuses on the month of June in our data set.

https://lh4.googleusercontent.com/2faiHq3yyKJ9YrlNYJ5opkCoOl5CggEOKu4Eot5Z9rjYNNGEOdgK3GKrGhCIEXJ_aqapnr9LQjrLKXjRZBobIlFZhfFn4i6HSpXfygB4ErgH2jNt3Lv95B5bEGOvhnJRkCFSxkn1EQ

Conclusions

Our conclusions are two-fold as they group into findings related to Cross-Platform Analysis in general and findings specific to the grexit dataset.

We explored the relations between different platforms and found that some platforms actively support the posting of content to other platforms while others do not. By mapping supported relations, we could see that what we call “secondary media platforms” (Youtube, Instagram, Tumblr) tend to feed content to the “primary platforms” (Twitter, Facebook).

Furthermore, we identified three ways to study the linkage between platforms (1) URLs appearing in posts linking to other platforms (2) platform-relevant hashtags (PRashtags), hashtags that carry information about other platforms, e.g. #youtube (3) source information attached to a post, indicating the source mobile client, e.g. instagram app.

When working with datasets from different platforms one has to keep in mind the bias introduced by the rate-limits of the respective APIs.

Findings specific to Grexit

Our cross-platform analysis revealed that instagram posts with hashtag grexit were mainly originating in Greece, while tweets primarily have been sent from western Europe. From the drastic difference in the geographic distribution we could conclude that the platform is tied to a geographical context, and it is indeed worth comparing different platforms when studying an issue.

We could also see that we found much more content on Instagram compared to Tumblr.

Future Work

In order to be able to derive generalizable statements about the linkage of the platforms, the three types of linkages should be assessed using a more comprehensive dataset. Furthermore, it would be interesting to explore how the linkages can be used to discover content across platforms, as an alternative to traditional hashtag and keyword search.





noxxg.jpgnovs4.jpg

Bibliography

Gerlitz, C. and Reider, B. (2013). “Mining one percent of Twitter: Collections, baselines, samples.” M/C Journal 16(2).

Knobel, M. and Lankshear, C. (2007). “Online memes, affinities, and cultural production” In C. Lankshear, M. Knobel, C. Bigum, & M. Peters (Eds.), A New Literacies Sampler 29 (199-227) New York: Peter Lang. ISBN 978-0-8204-9523-1.

Rogers, R. (2013). Digital methods. Cambridge, MA: MIT Press. ISBN 978-0-262-01883-8.

Shifman, L. (2013), Memes in a Digital World: Reconciling with a Conceptual Troublemaker. Journal of Computer-Mediated Communication, 18: 362–377. doi: 10.1111/jcc4.12013

Topic revision: r2 - 14 Jul 2015, micania
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback