The 2011 Tunisian uprising — the first domino in the series of mass popular
movements of the Arab Spring — witnessed the strategic utilization of
digital networks to bypass state security capacity and mobilize a
cross-sectional revolution. In Tunisia, Facebook and Twitter were used to
share updates about and coordinate the protests; the diffusion of ideas and
information on social media created a process of contagion that swept across
the whole region (Tudoroiu). In all corners of the world, the Tunisian
diaspora watched the Arab Spring and did not abstain from forming activist
networks. These online interactions allowed actors to virtually and
physically plan the fall of the regime — the Tunisian diaspora is an actor
that cannot be neglected.
While we do not absolutely embrace cyber-enthusiasm, which suggests that
social media was a “decisive tool in the coordination and execution” of the
Arab Spring movements, it is important to recognize how these digital spaces
“reduce[d] the cost of coordination” and facilitated the “dissemination of
information and literature” (Tudoroiu 2014). Even as traditional,
non-digital forms of political organization and discourse have established a
relatively healthy position in Tunisian society following the dictator’s
ouster, social media and its users maintain a crucial role in issue
formation and dialogue. In the wake of Tunisia’s July 2022 constitutional
referendum, the political waters are once again capricious, as scholars from
across the globe have earmarked this turn as a backslide to authoritarianism
(“Tunisia: Voters endorse new constitution, give President Saied more
power”). Further, with new media laws punishing the spread of ‘fake news’,
this issue of online activism is one that is topical (“Tunisia introduces
5-year jail term for 'fake news'”). Given the fraught Tunisian political
context and increased mobilization, our impetus is to explore the structure
and engagement of the Tunisian Twitter activist network.
Our digital inquiry thus poses a central question in relation to the
aforementioned context: How, and to what extent, do Twitter activists in
Tunisia shape and contribute to political issue formation on Twitter? In
order to explore this, however, it is important for us to delimit what we
define as activism. For the purpose of this inquiry, we consider activists
to be journalists, NGO members, scholars, or independents who consistently
use their Twitter platform to comment on topical political issues; we used
users’ bios to identify their professions. This exploration will delve into
the sub-questions of how Twitter activists operate within and perpetuate
echo chambers, (de)polarize online political camps, and, more broadly, shape
the political narrative on President Saied’s power consolidation. We aim to
unpack the question of how the Twitter algorithm facilitates the perception
of socio-political issues, specifically delving into the extent to which
Twitter dialogue has impacted the narrative of President Kais Saeid’s
regime. Moreover, we seek to evaluate the differing roles that activists can
take on the Twitter platform given their geographic proximity to the issue,
alongside how connected they are to other activists.
Literature Review and Preliminary Research
“Social media and activist communication” (Poell and van Dijck 2015)
This article takes the stance that social media usage is greatly informed by
the business models of social media corporations rather than user interests,
where retweeting, liking, following, friending, and algorithmic selection
mechanisms prioritize certain content. Poell and van Dijck focus on two main
defining attributes in the operation of activist communication: acceleration
and personalization. On one hand, acceleration is a positive force in
empowering activists to document events in real-time. However, the
instantaneous nature of these information updates means that social media
reporting has begun to mirror the criticism by which mainstream media is
usually defined: a dramatism of occurrences. Jodie Dean suggests that this
can largely be attributed to the fact that social media algorithms favor
more dramatic reporting due to the engagement it receives. This can be
observed on Twitter, with technological architectures such as ‘liking’,
‘retweeting’, hashtagging, and trending topic features which stimulate users
to spread breaking news (Poell and van Dijck). Twitter algorithms have shown
that clear spikes in topics are privileged over consistent engagement with a
certain issue. This may lead to activists to tailor their content to the
algorithm in order to reach the widest audience, potentially resulting in a
dramatized news feed.
Poell and van Dijck additionally discuss the role personalization plays in
online activism, providing further insight into how echo chambers, activist
networks, and trending topics may quickly materialize yet rapidly fade. They
cite that the speed at which social media platforms connect users can be
ascribed to the fact that social platforms “steer users towards personalized
connections, while at the same time introducing viral mechanisms in public
communication that produce moments of togetherness.” Parallels in this
debate can be drawn to Tufekci’s “Engineering the Public,” in which he
discusses the information asymmetry of data collection (Tufekci). Big data’s
process of categorization rather than pure individualization creates a
homogenization or convergence of preference; as Tufekci states: “all
categorization hides variation.” Echo chambers created online fundamentally
prevent varying ideas to challenge the dominant narrative. Poell and van
Dijck touch on how these techno-commercial models rely primarily on user
profiling and targeted advertising as a means of revenue generation.
Ultimately, monitoring trending topics “enables corporations to develop
real-time public sentiment tracking services” (Poell and van Dijck 533).
This calls into question the artificiality of performative activism, and how
corporate initiatives may undermine the goals of activists hoping to engage
with others in a sustainable, productive manner. Poell and van Dijk
perfectly capture the nature of online activism in a concluding statement:
“the interactions and interests that tie dispersed social media users
together to form protest movements, generating instant moments of
togetherness, inevitably dissolve when social platforms algorithmically
connect users to the next wave of trending topics” (534).
“The Tunisian diaspora: Between ‘digital riots’ and Web activism” (Teresa
Graziano 2012)
Tunisians abroad were the catalyst in communicating any and all matters
related to the 2011 protests. Graziano advances that “migrants mold
stratified social relationships across national boundaries and local
territories.” Diasporas played an important role in informing public
opinion, aided by cyberspace and the “radical shift in the extent, speed and
intensity of communication flows’ (Borkert, Cingolani, and Premazzi, 2009).
Graziano looks into how diasporas have capitalized on the use of the web
when Tunisia started experiencing unprecedented dynamics — or “irreversible
democratic transition.” “Conducting political conversations, exchanging
opinions, claiming for liberty and accessing credible information” was made
easy and this did not stop the diaspora from taking advantage to coordinate
political strategy. The diaspora established a “grassroots journalism” to
cover stories the mainstream media was reluctant to recognize. Networks of
cyber-activists were formed — in this paper's view, they are divided into
two categories: the ‘woke’ and the ‘enlightened’, where the woke assume to
know everything and the enlightened take a step back and privilege analysis,
distinguishing facts from interpretation.
Even after the revolution, Tunisian diaspora activists still use Twitter “to
assure up-to-the-minute news,” but more so to address political issues in a
deeper way. Tweeting completely replaced blogging and Tunisian “netizens”
have discovered the “highly attractive power of such tools to advocate their
legitimate democratic claims.” The “children of the revolution” are now 11
years older and are evermore concerned about their country. They continue to
discuss current affairs in Tunisia in great detail, and some have even
become references to major stakeholders in the region to advise on policy.
The author advances that the diasporas have “become the main connection
points for a widening network, the fundamental driving forces not only of
[...] remittances but above all [...] accessing the ‘information highways’.”
An example of an actor of the Tunisian diaspora “keeping an eye on Tunisia”
is 31-year-old Mohamed Dhia Hammami, a PhD student at Syracuse University.
Hammami grew up in Tunisia and participated in the 2011 Revolution. Today,
he observes Tunisia’s political scene from the US with tens of thousands of
followers — he is followed by Tunisians in Tunisia, and international
organizations, and is regularly solicited to speak about the current state
of political affairs. Current debates often ignore the fact that there is
not only one network in the Tunisian activist space. In Tunisia, the space
is made up of different activists who share one common sense of solidarity
for the fragility of the country. Overall, this author advances the
importance of diasporas, their contribution to their home societies, and
their tendency to form activist networks.
“Predicting Opinion Leaders in Twitter Activism Networks: The Case of the
Wisconsin Recall Election” (Weiai Xu et al.’s 2014)
Xu et al. delimit their research in explaining their focus on the Twitter
space’s engagement with a 2012 Wisconsin election event and the hashtag
#wirecall. Furthermore, the authors articulate the article’s aim of
understanding what elements contribute to individuals’ ‘opinion leadership,’
or “the ability to contribute information… and to lead others in
disseminating information;” they subsequently define ‘social connectivity’
and ‘issue involvement’ as necessary ideas in this empirical endeavor.
Moreover, while Xu et al. do not advance a particular normative claim
relating to Twitter activism, they stress the importance of “identify[ing]
influential [social media] users” in addition to studying institutional and
governmental instrumentalization of online network data. This conceptual
framework and individual-based analysis were helpful in guiding our
documentation of how digitally-influential activists interact with their
networks to shape discussion.
Our inquiry draws inspiration from the methodological tools employed by Xu
et al., as our research question required a similar nodal structure and
user-generated database. Xu et al. detail their methodological choices,
explaining that ‘social connectivity’ was measured by “betweenness
centrality,” or “the user’s strategic location in reaching all other users
in the network;” and that ‘user involvement’ was quantified through
“revealed political information on Twitter profiles,” “geographical
proximity to a given political event,” and the “contribution of engaging
Tweets;” and that ‘opinion leadership’ was assessed through the “number of
times [a] user’s tweets are retweeted.” Moreover, Xu et al. used network
analysis tools such as NodeXL as well as regression analysis to represent
and test the collected data visually. This precise concept-to-measure and
statistical approach was useful in extracting quantitative points from
user-generated Tweets and profile data in order to empirically test our
hypotheses.
While Xu et al.’s research concentrates on the nuances behind individual
‘opinion leadership,’ it is positioned within the broader political
polarization and echo-chamber formation debates. Understanding how
individuals augment their online social capital and clout as an
information-transmitter lends insight into why certain figures are
particularly polarizing or influential. A 2021 NYU report, “Fueling the
Fire: How Social Media Intensifies U.S. Political Polarization,” asserts how
“social media ‘has become a powerful accelerant’” of political hostility.
The report highlights the significance of the ‘Trump factor’ — how former
President Donald Trump and his followers “pursued an unprecedented social
media campaign aimed at provoking us-vs-them hatred.” Although Trump is not
an ordinary activist, his use of social media demonstrates the extent to
which influential people can use charged language to entrench political
sectarianism. Our paper employs Tweet analysis methods in order to consider
how Tunisian activists expand or contract their follower bases. A Science
article, “Political Sectarianism in America,” further explores Twitter and
Facebook’s “popularity-based algorithms that tailor content to maximize user
engagement, increasing sectarianism within homogeneous networks” (Bail et
al. 2020). Building upon Xu et al.’s discussion of how “engaging Tweets”
contribute to ‘user involvement’ and thus higher ‘opinion leadership,’ Bail
et al. explore how language works in tandem with social media algorithms to
fuel the dissemination of highly partisan messages. Our inquiry draws on
these findings in our study of how Tunisian Twitter activists influence echo
chambers through the choice of language.
Xu et al. found that “betweenness centrality was positively related to the
[user’s] number of” retweets and that “issue involvement based on engaging
tweets… positively predicted” retweets. These results are context-specific,
however, they can nonetheless be applied to an exploration of how Tunisian
Twitter activists engage with their audiences, be they diverse or
homogenous. Xu et al.’s acknowledge that their work is limited by their
study of only people living in Wisconsin, thereby discounting high ‘opinion
leadership’ individuals elsewhere in the U.S. Our inquiry seeks to consider
Twitter activists based outside of the capital, Tunis, for more sound
results.
Hypotheses
It seems to go without saying that those activists who are well-connected
will have the greatest outreach. We thus employ the notion of social capital
in the digital realm, assuming that influence within these political
discussions is dependent on how connected one is to other users. In saying
this, we hypothesize that there is a positive relationship between those
activists who utilize the most inflammatory language and those who we
categorized as being in the ‘center’ of the network. This is on the basis
that provocative language will elicit emotional responses from more users,
prompting retweets, replies, and likes. Given our knowledge of the Twitter
algorithm discussed in our preliminary research, spikes in engagement are
favored in determining the outreach of user activity. Those tweets that use
emotive language would engage a wider audience, consequently building up an
activist's platform. Another hypothesis that can be drawn from this
conclusion is that echo chambers are created through this process of
support. As different activists engage and align themselves with others,
through retweeting and liking, it is likely that those who share similar
political opinions on given topics will be connected. Our final hypothesis
is that those activists who would be considered part of the Tunisian
diaspora would have higher Twitter engagement, most likely because of the
fewer restraints placed upon their freedom of speech when living outside of
Tunisia. Kais Saied’s power consolidation has brought stricter media laws,
thereby impacting the likelihood that those living within Tunisia use their
Twitter platforms to propagate anti-regime views.
Methodology
This paper addresses a topic that is qualitative at its core and therefore
we thought it best to adopt a netnographic strategy to observe the
phenomena that take place within the realm of Twitter connectivity.
Nonetheless, we sought to include some quantitative data to supplement
some of our more subjective findings. To start, we searched for 16
prominent activists that operate both within and outside of Tunisia; we
used Tweet Deck to narrow down our search based on a variety of keywords
such as ‘Kais Saied’ or ‘KS’, ‘Coup’, ‘July Takeover’, ‘Constitution’, and
‘تونس’ (Tunisia). Often a combination of these key phrases would be used
or by the same prominent activists; however, if they did not, we used
broader terms, such as ‘President’. As discussed, we categorized activists
as those individuals in academia, humanitarian work, journalism, and
political analysts who actively use their platform to comment on a
particular aspect of the Tunisian political context. Once the 16 activists
were selected we decided to look at social connectivity and networks, as
discussed by Xu et. al, by using Gephi, a nodal mapping software. We
manually went through which activists followed one another or not, using
these relationships to construct nodal maps which visualize the
intra-network relationships, specifically mutual connections (where both
users follow one another) and one-sided connections (outgoing followings;
those users which the activist in question follows). By assembling these
networks, we were able to discern which users were in the ‘periphery’ and
which were ‘central’; this allowed us to understand how proximity to other
users impacted their influence within the network, or as posited by Xu et
al, their “betweenness centrality.”
Furthermore, we analyzed specific Tweets made by the 16 activists on
Tunisian President Kais Said’s consolidation of power. Utilizing Tweet
Deck, we repeated our keyword search alongside important political dates
such as the Saied’s suspension of parliament on the 25th of July 2021. We
chose five Tweets per activist in order to analyze engagement through
retweets (RTs), likes, and the use of inflammatory language. We
categorized Tweets into two groups: (1) those which expressed a clear
opinion (using biased language) and (2) those which utilized provocative
language. From this we were able to find the average number of retweets
and likes for both group 1 and group 2, helping us evaluate the extent to
which inflammatory Tweets garner greater engagement. Further, we examined
the number of group 1 and 2 Tweets used by both the central and peripheral
categories to understand the relationship between inflammatory language
and users’ “betweenness centrality.”
Limitations
Despite our attempts to use a robust, multi-faceted methodology, there are several limitations to our exploration. The basis for which we defined an activist was largely subjective; although we provided a definition in our introduction, we feel that if we employed a more systematic approach, we could have used a greater number of users within a specific follower range. This would have made the data more representative. Twitter’s rejection of our API license limited the breadth of our research, making it more difficult to systematically generate larger quantities of data. Hence, our methodology relied heavily on Tweet Deck — a temperamental site with some minor glitches that did not always provide the most telling results. In hindsight, we also chose to use keywords pertinent to the case study of Kais Saied’s political regime. This limited our ability to find relevant tweets by our selected activists, as many of the activists used their platforms to disseminate other topics, such as gender equality or environmental justice. We also recognize that our analysis of language is another aspect of this exploration which is subjective. However, it can be argued that all language to a certain degree is up to interpretation; we feel that our justifications as to why we categorized certain Tweets as inflammatory presents some legitimacy in our findings. Nonetheless, it is important for us to also recognize that the nature of a netnographic approach relies heavily on observation and a dissection of phenomena, discussions, and expressions of language. The approach taken is thus inherently subjective.
Analysis
Figure 1: Nodal Map Analysis
These nodal maps visually represent the ‘betweenness centrality’ (Xu. et al) of different Twitter activists, illustrating the relationship between each user through the mutual and one-sided connections. Those nodes on the periphery of the above network represent those users who have a greater number of one-sided connections (where they follow other activists) compared to mutual connections, and a lower total number of connections in general. Conversely, central users tend to have more mutual connections (users that follow them back) compared to one-sided connections. What is interesting, however, is that the average number of followers for the central users is 43,000, whereas for the periphery it is 109,380. This almost goes against the idea that the central users would have more people following them. It can be assumed that there perhaps would be a correlation between the activists who have the most followers and how connected they are. However, this nodal map illustrates connectivity rather than sheer follower count. Therefore, it can be said that mutual connectivity is favored over one-sided relationships in determining how central a user is. This is evidenced by the fact that the three users with the highest mutual connectivity are central and the five users with the lowest mutual connectivity are on the periphery. These connections could be impacted by the fact that central activists seem to be academics outside of Tunisia or people in prestigious research or political positions, such as Youssef Cherif, the Director of Columbia Global Centers Tunis. Therefore, it makes sense that they are the most connected to other activists even if they are not the most followed given the nature of their work to be engaged with a variety of actors. The below graph depicts the relationship between users’ mutuals and people followed.
Figure 2: Followed vs. Mutual chart
Strongest networks
@Selim_ → 30 total connections (mutual + one-sided); 1.00 following/mutuals ratio (i.e 100% of the people @Selim_ follows also follow him)
@FadilAliriza → 25 total connections; 0.92 following/mutuals ratio
Weakest networks @Adelazouni → 10 total connections; 0.5 following/mutuals ratio
@HaythemElMekki → 15 total connections; 0.56 following/mutuals ratio
@HaythemElMekki → 15 total connections; 0.56 following/mutuals ratio
19 Tweet Analyses – Kais Saied
The Tweet analysis represented below includes commentary on the language choices made by several central and peripheral activists in explicit or indirect reference to President Kais Saied’s actions. For reference, the 25th of July, 2021 is the date that President Kais Saied suspended the Parliament by invoking Article 80 of the Tunisian Constitution. The comments provided on @Selim_’s Tweets demonstrate how charged, inflammatory language and declarative rather than generalized statements increased engagement (likes and RTs) with his Tweets. Moreover, @Sarah_bh’s Tweets tended to receive more engagement when she expressed her views as opposed to providing neutral or confused commentary. @MedDhiaH’s most engaging Tweets were those with the most politicized, inflammatory, and sarcastic messages. Interestingly, @OuiemCh’s Tweet garnering the most attention incorporated a reference to personal experience and emotion, indicating that sensitivity may sometimes be prioritized over strictly political commentary. @Tmegrisi’s Tweets underline how references to the faults of an individual (President Saied, in this case) are more engaging than those which exhibit a general frustration with the system. Finally, @SofiaNaceur’s Tweets illustrate how exigent language creates more engagement than calmer, more analytical language; this may be due to the fact that Tweets made directly after a shocking event evoke more immediate fear and uncertainty.
With our two categories of language — (1) those which embodied a clear opinion, and (2) those which utilized provocative language — we found the average number of retweets and likes for both group 1 and group 2.
This illustrates that more inflammatory Tweets (group 2) produce more engagement. We do recognise that the margin by which the group 2 averages are greater is not too significant, however if this were to be observed on a larger scale, we would still expect the same phenomenon to occur.
Further, as we hypothesized, there is a positive correlation that can be observed between those activists who utilize the most inflammatory language and those activists which occupy the center of the nodal network. Out of a pool of 35 selected tweets, we found the following:
Central [18 total tweets] → 6 inflammatory, 12 opinion Peripheral [17 total tweets] → 3 inflammatory, 14 opinion
Thus, we can extrapolate that given group 1 tweets garnered more engagement and because the center had more inflammatory tweets, that center activists tend to employ more provocative language. This aids in their overall influence within the nodal network.
This section seeks to reflect on the second hypothesis: That echo chambers would form within the Twitter activist community under scrutiny. The table pictured above (left) portrays the intra-network interactions of the two activists (@Selim_ and @FadilAliriza) with the strongest networks (the highest sum of mutual and one-sided connections and following-to-mutual connection ratio closest to 1.0) and the two weakest networks (@Adelazouni and @HaythemElMekki). The interactions were taken between 20 July 2021 and 11 October 2021 to ensure an adequate sample size and are exhaustive, meaning that no other intra-network interactions were observed within this time frame. All of the retweets by both activists with strong and weak networks were supportive of the respective author’s message, most notably because no additional comments were added — a sign of endorsement or affirmation. This arguably reflects the nature of an ‘echo chamber,’ or the process of “group polarization [...that] occurs in intensive deliberations by small groups of homogenous individuals on topics of high relevance and concern,” (Kitchens et al., 2020) such as Kais Saied’s ‘constitutional coup’. Interestingly, however, is the fact that those with strong networks did not confine themselves to engagement with the center of the network; the majority of @Selim_ and @FadilAliriza’s interactions were with peripheral activists. On the contrary, the two activists with the weakest networks interacted most frequently with central activists’ Tweets. It can thus be concluded that central activists’ information input is more diverse than that of peripheral activists in this network, indicating that peripheral activists may perpetuate their position in the outer sphere by engaging solely with the center. In other words, peripheral activists do not cast their networks as wide as activists in the center of the nodal map do.
Hypothesis 3: The Tunisian Diaspora has higher Twitter engagement
In this specific nodal network, it was observed that those activists located within Tunisia were more prominent in both the center and the periphery. This effectively nullifies our hypothesis. It could be argued that those inside Tunisia have a proximity to the causes they Tweet about and therefore could amass greater engagement than those geographically removed
However, it is important to note that our selected activists were not balanced; there was not an equal number of users from the diaspora and those within Tunisia. In saying this, the fact that a significant percentage of the central activists represented almost half of the total share conveys that the diaspora remains an influential entity in Twitter discourse. This is solidified by our previous analysis concerning the sway of central activists in their ability to engage users. This links again to the nature of the central activists. If they are studying Tunisia as part of their professions as researchers, academics, humanitarian workers, and political analysts, it is arguably a requirement that they would have to be engaged with Tunisian issues to a greater extent. Therefore, they would exhibit greater connectivity within the nodal framework to other actors as a means to stay informed.
Conclusion
This digital exploration attempted to shed light upon the extent to which Twitter activists in Tunisia shape and contribute to political issue formation on Twitter. With respect to our three hypotheses, several conclusions can be drawn. First, this project has demonstrated that mutual connections are favored over one-sided connections in determining centrality in an online network. Moreover, it has been shown through a Tweet and average analysis that inflammatory, provocative, targeted, and declarative language garner more engagement than more neutral, generalized Tweets. The second hypothesis concerning the creation of echo chambers in the Twitter network under analysis was supported in that members of the network interacted with each other, primarily through retweets, in an affirmative manner, thereby indicating the maintenance of a shared political narrative. Interestingly, peripheral activists were shown to preserve their position in the network by engaging primarily with central activists, while central activists perpetuated their role in the core by engaging with a broad reach of peripheral nodes. Finally, it has been shown that in the 16-member activist network, Tunisia-based users were proportionally more prevalent in both the center and periphery; however, it is important to stress that almost half of the activists analyzed are from the diaspora, highlighting their relevance in the broader discussion of Twitter political issue-formation. Despite the methodological and logistical challenges faced in the realization of this digital inquiry, the findings presented offer a preliminary glance into how the Tunisian activist ‘Twitter-sphere’ operates and insight into how Twitter is employed in issue-formation in a non-Western context.
.png?v=1669383113868)
TWITISIA
