Mainstream and Alternative Narratives in the Wake of Gun Shootings
Lisa Grobelscheg
1,2
, Ema Ku
ˇ
sen
3
and Mark Strembeck
1,4,5
1
Institute of Information Systems and New Media, Vienna University of Economics and Business, Vienna, Austria
2
FH CAMPUS 02, University of Applied Sciences, Graz, Austria
3
Faculty of Informatics, University of Vienna, Vienna, Austria
4
Secure Business Austria (SBA), Vienna, Austria
5
Complexity Science Hub (CSH), Vienna, Austria
Keywords:
Mass Shootings, Narratives, Terror Attacks, Twitter.
Abstract:
In this paper, we explore narratives that emerge in the Twitter discourse during high stakes, low probability
events. In particular, we analyze 7.4 million tweets related to four shooting events in the United States of
America to identify differences that arise in the semantic structure and message diffusion of mainstream nar-
ratives on the one hand and alternative narratives on the other. Our findings indicate that alternative narratives
introduce keyterms that have little to no connection to the respective shooting itself and that their diffusion
patterns similar to those of mainstream narratives. Moreover, we found empirical evidence of alternative nar-
ratives, such as false flag accusations, that appear across different events and persist in the Twitter-sphere over
an extensive period of time.
1 INTRODUCTION
If current news about a situation are in high demand,
e.g. during a crisis event (Stieglitz et al., 2018), in-
formation shared via social media can have a strong
impact on the users being exposed to it. It stands to
reason that certain narratives emerging in the wake
of a crisis event might even cause trauma or emo-
tional pain in users engaging in public discourse (see,
e.g., (Goodwin et al., 2018; Ku
ˇ
sen and Strembeck,
2021)). In this paper, we define a narrative as a a
set of topic-wise interconnected messages that have
been sent/posted via a social media platform (see
also (Cunliffe et al., 2004; Weick, 1995; Grobelscheg
et al., 2022)).
Van Proojen and van Dijk (van Prooijen and van
Dijk, 2014) indicated that two factors correlate with
the probability of narratives being connected to con-
spiracy theories in the public discourse. Firstly, the
scale of consequences caused by an event and sec-
ondly, the degree of perspective-taking with the vic-
tim group of an event. Based on these findings, we
hypothesize a high potential of narratives describing
conspiracy theories in social media discourse in the
wake of crisis events. In this paper, we focus our
work on the analysis of prevalent narratives related
to gun shootings in the United States. Gun shootings
affect all classes of the population in the U.S. as most
attacks hit “soft targets”, such as concert venues or
schools (see, e.g., (Ku
ˇ
sen and Strembeck, 2021)).
Following the intuition of (Starbird, 2017a) and
(Nied et al., 2017) we built on the concept of “alter-
native narratives”. As opposed to the notion of con-
spiracy theories and fake news, which inherently in-
clude false information, an alternative narrative might
also convey truthful information. A distinctive feature
of an alternative narrative is the contradiction of the
mainstream view of an event or person (Wang et al.,
2022). Building on this concept, we analyze alterna-
tive and mainstream narratives arising in social media
in the wake of gun shootings in the U.S. To this end,
we investigated semantic structures and hashtag asso-
ciations of narratives as well as user activity. In par-
ticular, we aimed to discover differences in the narra-
tive structure of alternative and mainstream narratives
in order to find similarities in alternative narratives
across multiple events.
The remainder of this paper is organized as fol-
lows. Section 2 covers related work on the construc-
tion of narratives in social media and the emergence
of alternative narratives. Section 3 briefly describes
the events considered in this paper and Section 4 out-
lines our research method and research questions. We
present our results in Section 5 and discuss them in 6.
Grobelscheg, L., Kušen, E. and Strembeck, M.
Mainstream and Alternative Narratives in the Wake of Gun Shootings.
DOI: 10.5220/0011972800003485
In Proceedings of the 8th International Conference on Complexity, Future Information Systems and Risk (COMPLEXIS 2023), pages 17-26
ISBN: 978-989-758-644-6; ISSN: 2184-5034
Copyright
c
2023 by SCITEPRESS Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)
17
The final section 7 concludes our paper and provides
an outlook for future research directions.
2 RELATED WORK
2.1 Narratives in Social Media
Narratives in social media have been investigated
from various perspectives. Li et al. used the narrative
dimensions proposed by Fisher (Fisher, 1984) to anal-
yse user health information adoption (Li et al., 2019).
They found that information scoring high in narrative
fidelity (how trustworthy a story sounds compared
to the reader’s own experiences) as well as narrative
probability (how coherent a story appears) is more
likely to be adopted by users in social media. As the
credibility of information is crucial during and after
crisis events, it has been the subject of numerous stud-
ies. For example, Hardy and Miller analyzed Twitter
messages after a shooting at a Nightclub in Orlando,
Florida in 2016 (Hardy and Miller, 2022). They found
two main narrative themes among post-crisis narra-
tives, namely unification and division. Both themes
serve a different purpose. While division provokes
future action, unification is a means of coping.
2.2 Emergence of Alternative
Narratives in Social Media
Sometimes, the narratives in social media on the one
hand and classical mass media outlets (such as news-
papers, radio or TV channels) on the other diverge or
even contradict each other. For example, while clas-
sical mass media outlets suggested an Islamic back-
ground to the 2011 Norway attacks, the discourse on
Twitter condemned its inaccurate reporting, framing a
competing narrative (Eriksson, 2016).
Alternative narratives might especially appear in
the wake of a crisis event. By analyzing social me-
dia narratives in the wake of the Boston Marathon
Bombing of 2013, claims of the event being a “false
flag” appeared online within minutes of the blasts and
kept spreading for months after the event (Nied et al.,
2017). In (Starbird, 2017b), Starbird investigates the
propagation of alternative narratives through a me-
dia ecosystem. Starbird defines alternative narratives
as narratives offering an explanation of a man-made
crisis event opposing the narrative found in classical
mass media.
Another approach used Twitter data in the af-
termath of the 2015 measles outbreak to create
a narrative structure (Radzikowski et al., 2016).
Radzikowski et al. use spatial data, important terms,
as well as special types of communication (such as
retweets) to identify a narrative of anti-vaxers. Ad-
ditionally, they applied clustering on the hashtags
used on the tweets to identify subgroups of narratives.
In addition, they differentiate between “influencers”
posting a message and “amplifiers” retweeting it. In
(Klein et al., 2019), Klein et al. investigated social and
linguistic features of users engaged in a Reddit con-
spiracy forum and found distinctive features in user
behavior even before users post in conspiracy forums.
Fong et al. (Fong et al., 2021) also analyzed user
data (e.g., followers and tweets) of users in a Reddit
conspiracy forum. They conducted a lexical analy-
sis using LIWC (Tausczik and Pennebaker, 2010) and
found that users proclaiming conspiracy theories in-
clude more negative emotions (e.g., anger) in their
tweets. Also words related to power, death, and reli-
gion are common. Furthermore they express stronger
orientation to the past and out- and in-group language
(“us vs. them”).
3 EVENTS OF STUDY
2017 Las Vegas Shooting. On October 1, 2017, a
mass shooting at the Route 91 Harvest music festival
in Las Vegas, Nevada killed 60 people and wounded
at least 416 others. The subsequent panic resulted in
additional 451 injuries
1
.
2018 San Bruno Shooting. On April 3, 2018, Nasim
Najafi Aghdam fired shots at the YouTube headquar-
ters, injuring three individuals before committing sui-
cide
2
.
2018 Santa Fe School Shooting. On May 18, 2018,
eight students and two teachers were shot dead and 13
others wounded in a school shooting at Santa Fe high
school in Santa Fe, Texas
3
.
2019 El Paso Shooting. On August 3, 2019, Patrick
Wood Crusius fired shots at a Walmart in El Paso,
Texas, killing 23 people and injuring 23 others
4
.
1
see, e.g., https://www.reviewjournal.com/crime/shooti
ngs/las-vegas-woman-becomes-60th-victim-of-october-2
017-mass-shooting-2123456/
2
see, e.g., https://edition.cnn.com/2018/04/04/us/yout
ube-hq-shooting/index.html
3
see, e.g., https://www.click2houston.com/news/local/2
022/05/18/santa-fe-high-school-shooting-4-years-later-e
vents-planned-to-mark-anniversary/
4
see, e.g., https://www.buzzfeednews.com/article/mar
yanngeorgantopoulos/el-paso-shooting-white-supremacist
-terror-attack-victims
COMPLEXIS 2023 - 8th International Conference on Complexity, Future Information Systems and Risk
18
Days after the event
mainstream
alternative
Percent of daily Tweets
(a) Percentage of daily tweets for El Paso.
Narrative
Mainstream
Alternative
Days after the event
Percent of daily Tweets
mainstream
alternative
(b) Percentage of daily tweets for Las Vegas.
Days after the event
Percent of daily Tweets
mainstream
alternative
(c) Percentage of daily tweets for Santa Fe.
Days after the event
Percent of daily Tweets
mainstream
alternative
(d) Percentage of daily tweets for San Bruno.
Figure 1: Proportions of daily Tweets for alternative and mainstream narratives for days 1 to n for each event.
4 RESEARCH APPROACH
Our paper is guided by the following research ques-
tions.
RQ1: What is the Semantic Structure of Alterna-
tive Narratives? After a manual labeling procedure
of tweets into tweets conveying mainstream and al-
ternative narratives (see Section 4.1), we investigated
the language used in both types of narratives. We ap-
plied text mining techniques, such as part-of-speech
tagging (POS) and word co-occurrence analysis to
identify semantic patterns. Furthermore, we created
a hashtag co-occurrence network to analyse content
communities.
RQ2: How Do Users Engage in Alternative Nar-
ratives Across Multiple Events? To compare char-
acteristics in terms of diffusion and user engagement,
we analyzed retweet activity and user behavior across
multiple data-sets.
4.1 Procedure
Data Collection. We used Twitter’s search API to
obtain tweets related to four mass shooting events in
the U.S. The hashtags and keyterms used for the data
respective extraction procedure, as well as the obser-
vation periods are presented in Table 1. After the re-
moval of duplicate entries, our data collection counted
over 7.4 million tweets.
Defining Mainstream and Alternative Narratives.
To identify and distinguish between mainstream and
alternative narratives, we used ve classical media
outlets which, according to a Pew Reserach Cen-
ter report (Shearer and Mitchell, 2021), are consid-
ered mainstream media by a large proportion of U.S.
American citizens. This list includes: i) ABC News,
ii) CNN, iii) MSNBC, iv) FoxNews and v) the Wall
Street Journal.
The filtering procedure for news articles in these
mainstream media outlets included the location of the
shooting and a descriptive word (e.g., Santa Fe school
shooting, El Paso shooting) as search terms for the
above-mentioned online news portals. In our study,
we considered articles published up to three months
after a shooting event. In total, we manually inspected
130 related news articles.
This procedure resulted in eight narratives, which
were subsequently categorized into three types of
Mainstream and Alternative Narratives in the Wake of Gun Shootings
19
Table 1: Basic information about data-sets.
Location Observation
period
Victims Users Tweets Search terms
Dead Wounded
Las Vegas (NV) 2-14 Octo-
ber 2017
60 867 1,394,070 3,436,187 #lasvegasattack, #LasVegasMassShoot-
ing, #LasVegasShooter, #LasVeg-
asStrong, #PrayforLasVegas, #Pray-
ingForLasVegas, shooting Las Vegas,
#StephenPaddock, #StephenPaddock-
IsATerrorist, #VegasStrong
San Bruno (CA) 4-10 April
2018
0 3 312,206 648,501 #YouTubeHQShooting, #NasimNajafi-
Aghdam, #SanBurnoShooting, #shoot-
ingYouTube, #YouTubeHQTragedy,
#youtubeShooter, #YouTubeStrong,
#YouTubeShooting,
Santa Fe (TX) 18-25 May
2018
10 13 458,644 967,674 #DimitriosPagourtzis, #SantaFe,
#SantaFeGunControl, #SantaFe-
GunControlNow, #SantaFeViolence,
#SantaFeHighschool, #SantaFeStrong,
#SantaFeShooting,
El Paso (TX) 3-18 Au-
gust 2019
23 23 939,940 2,307,577 #ElPaso, #elpassoshooter, #ElPa-
soshooting, #ElPasostrong, #ElPa-
soterroristattack, #massshooting,
#PrayersforElPaso, #PrayforElPaso,
#walmartshooting, #domesticterrorism
mainstream narratives: pro-gun narratives, anti-gun
narratives, and general narratives, as summarized in
Table 2. For each news article, we derived the most
prevalent narrative conveyed in that very article. This
implies that an article might also include messages of
(for example) unity even though it is labeled as blam-
ing mental health issues.
The narrative of blaming mental health issues es-
pecially includes mentions of wrong or insufficient
treatment of mental health disorders and illnesses, as
well as possible side effects of medication against it
(e.g., Ritalin). In messages related to the Santa Fe gun
shooting we found accusations towards (“young”)
people’s attitude towards the value of life (“life is
worth nothing”). In this light, abortion rights, video
games, and porn consumption were mentioned as the
main culprits for the act of shooting. Since many gun
shootings are directed at “soft targets” such as schools
or entertainment venues, pro-gun narratives occasion-
ally blame the security measures of these structures.
Common demands include more weapons for teach-
ers, less entry points and more security personnel. For
all events, pro- and anti-gun advocates formulated ac-
cusations for the opposite side. Anti-gun narratives
include demands for stricter gun laws or less power
for the National Rifle Association (NRA). General
narratives include appeals for hope and unity.
Table 2: Narratives in mainstream media.
MAINSTREAM NARRATIVE EVENT
Pro-gun narratives
blaming mental health issues, medication
against it (e.g. Ritalin)
Santa Fe,
Las Vegas
blaming attitude towards life, abortion,
porn, video games
Santa Fe
blaming vulnerability of soft targets El Paso,
Santa Fe
blaming opposing politicians, e.g. repub-
licans blaming democrats
all events
Anti-gun narratives
blaming current gun laws all events
blaming influence of National Rifle Asso-
ciation (NRA)
Santa Fe
blaming opposing politicians, e.g.
democrats blaming republicans
all events
General narratives
stop the hate, unite as one nation Las Vegas
Narrative Discovery in Our Collection. We build on
the conclusions of (Zappavigna, 2015) who suggest
that hashtags are widely used as topic-markers and
COMPLEXIS 2023 - 8th International Conference on Complexity, Future Information Systems and Risk
20
possess the ability to add structural and interpersonal
meaning. In our analysis, we used hashtags to distin-
guish between alternative narratives and mainstream
media narratives. For this task, we deployed two an-
notators who manually examined the list of hashtags
and labelled them as either “mainstream” or “alter-
native”. The annotators were guided by the list of
mainstream narratives that resulted from the news ar-
ticle inspection described above. For some hashtags,
the annotators inspected a random sample of tweets
containing the corresponding hashtags to decide on
classification of that label. This procedure resulted
in a substantial inter-rater agreement (Cohen Kappa
79.35%).
After the labels were revised and any discrepan-
cies between the two annotators resolved, the propor-
tion of alternative narratives in the data-set was 2.2%
for the Las Vegas shooting, 4.6% for YouTube HQ
shooting, 0.9% for El Paso Walmart shooting, and
0.1% for the Santa Fe school shooting.
Semantic and User Analysis. For our semantic anal-
ysis, we used the UDPipe R package (Straka and
Strakov
´
a, 2017) to carry out Part-of-Speech-Tagging
(POS), word frequency, and a co-occurrence analysis.
Furthermore, we created a hashtag co-occurrence net-
work and applied the Louvain-clustering algorithm to
obtain themes in the hashtags of the four events.
5 RESULTS
Semantic Analysis. First, we investigated the corre-
lation between proportional word frequencies among
alternative and mainstream narrative tweets. Pear-
son’s product-moment correlation sample estimate is
between 0.2347 for Santa Fe and 0.3051 for El Paso
which indicates a low correlation of word proportion
between the two narrative subsets. The proportion
of original messages (excluding retweets) among all
messages accounts to 17.0415% for alternative and
18.4892% for mainstream messages. We then car-
ried out a temporal analysis of alternative vs. main-
stream narratives in order to identify differences in
their prevalence. Figure 1 depicts the temporal (pro-
portional) prevalence of alternative and mainstream
narratives. The analysis indicated a time-lag in preva-
lence of alternative narratives which is intuitive as al-
ternative narratives do not include news updates dur-
ing the event. Across data-sets, we found peaks of
alternative tweets on certain days after an event. No
obvious connection to real-world incidents could be
identified for the peaks in alternative narratives.
As described above, we applied part-of-speech
tagging to investigate structural differences. Figure
2 depicts the proportional difference in prevalence
of universal part of speech categories between al-
ternative and mainstream narrative. For example,
in alternative narratives, the proportion of verbs is
1.59% higher than in mainstream narratives. More-
over, mainstream narratives include more pronouns,
prepositions (such as in, to, during) and adjectives
compared to alternative narratives. Interestingly, the
proportion of undefined tokens “X” was higher for al-
ternative narratives. The most frequent words in the
“X” category were: qanon, falseflag, george, good-
man, jason, webb, maga and wwg1wga (for our anal-
ysis, we excluded stop words such as lasvegasshoot-
ing).
To investigate the use of entities in alternative and
mainstream narratives, we analyzed the top 15 nouns
of each corpus and aggregated them in Figure 3. As
expected, the top alternative nouns were populated
by words associated with conspiracy theories, such
as ”wwg1wgaworldwide”, which stands for ”where
we go 1, we go all worldwide” and is used by Qanon
followers. “Fakenews”, “fakenewsmedia” and “msm”
(mainstream media) on the other hand accused main-
stream media of spreading misinformation to support
a hidden agenda. We also identified attempts to de-
clare the event as a “falseflag” attack carried out by
the government itself to cover-up for other actions to-
wards alleged secret goals.
Mainstream word analysis included references to
other shootings, labels such as “domestic terrorism”
and entities directly associated with the events (e.g.,
“walmart”, “victims” or “shootings”).
Furthermore, we carried out a co-occurrence anal-
ysis for each narrative and data-set. Co-occurrences
cover combinations of nouns and adjectives. Fig-
ure 4 depicts the 30 most prevalent combinations per
tweet. The number of items in each Figure varies, as
some terms might be included in multiple prevalent
combinations (e.g., for mainstream narratives for San
Bruno: gun - woman, gun - state).
When comparing word networks of alternative
narratives with mainstream narratives, positive words
that can be found in mainstream narratives (e.g.,
“prayer”, “thought”, “love”) cannot be found in the
word networks resulting from alternative narratives.
Also, across all events, we found alternative narra-
tives to be connected with conspiracy theories, such
as Qanon. Prevalent word combinations were, for ex-
ample, “deep state”, “qanon”, “george soros”. Inter-
estingly, alternative narratives often included names
of well-known individuals who were not directly con-
nected to an event itself (e.g., George Soros, Jason
Webb). This could not be observed for mainstream
narratives. We often found negative references to
Mainstream and Alternative Narratives in the Wake of Gun Shootings
21
alternative
mainstream
average proportion
Figure 2: Difference in UPos categories between mainstream and alternative narrative.
frequency in percent
wwg1wgaworldwide
(a) Top 15 nouns in alternative narratives proportional to all
words per data-set.
frequency in percent
(b) Top 15 nouns in mainstream narratives proportional
to all words per data-set.
Figure 3: Top 15 nouns for across all events.
“the” media (used for referring to mainstream media
outlets) in alternative narratives. For example, “fake-
newsmedia” as a central connection phrase in the El
Paso data-set.
Users Engaging in Alternative Narratives. Our user
analysis was led by two main questions, namely: “Do
certain users exclusively participate in the spreading
of alternative messages?” and “How many of the
users who are spreading alternative narratives partic-
ipate in multiple events?”. To compare our numbers,
we also analyzed users spreading mainstream narra-
tives with respect to the above stated questions.
Most users exclusively contribute to mainstream
narratives (on average 97.51% of all users across all
data-sets), whereas 1.57% (numbers rounded) engage
in alternative narratives only, and 1.91% of all users
contribute to both types of narratives. The number
of messages sent per user is slightly higher for main-
stream narratives (1.786 messages/user) as opposed to
alternative narratives (1.317 messages/user). 2.49%
of all users contributed to an alternative narrative at
least once, 7.32% of these users contributed to nar-
ratives related to multiple events. For users who en-
gaged in mainstream narratives at least once 10% con-
tributed to narratives related to multiple events. In
contrast, only 1.57% of the users who exclusively par-
ticipated in the spread of alternative narratives con-
tributed to multiple events. Therefore, we conclude
no higher repeated engagement for alternative narra-
tives as compared to mainstream narratives.
Hashtag Co-Occurrence Networks. We created net-
works from hashtag co-occurrence and applied the
COMPLEXIS 2023 - 8th International Conference on Complexity, Future Information Systems and Risk
22
ies
(a) Word network for El Paso - alternative narratives.
(b) Word network for El Paso - mainstream narratives.
kenews
(c) Word network for Las Vegas - alternative narratives.
prayforla as
veg
(d) Word network for Las Vegas - mainstream narratives.
(e) Word network for Santa Fe - alternative narratives.
(f) Word network for Santa Fe - mainstream narratives.
qanon
(g) Word network for San Bruno - alternative narratives.
people
(h) Word network for San Bruno - mainstream narratives.
Figure 4: Word co-occurence networks for each event and narrative.
Louvain-clustering algorithm to obtain communities,
or hashtag themes. These networks include the most
popular hashtags for each network
5
. For this analysis,
5
We took the number of appearances of a hashtag and
used the 0.9995 percentile for El Paso and Las Vegas, 0.998
for Santa Fe and 0.995 for San Bruno
we did not distinguish between alternative and main-
stream hashtags, as our aim was to investigate if any
alternative themes would arise.
Figure 5 depicts hashtag co-occurrence networks
for all events. For the attack in Las Vegas, we found
one hashtag theme focusing on hope and prayer (e.g.,
Mainstream and Alternative Narratives in the Wake of Gun Shootings
23
#prayingforvegas, #prayforvegas). The other two
themes are connected to gun control and mixed dis-
cussions (see green theme). For El Paso, we find
mainly mixed clusters, dealing with references to
other attacks (Chicago, Dayton), gun control and the
classification of the event as “domestic terrorism”.
Our analysis indicates a strong connection (see edge
width in Figure 5a) between #domesticterrorism and
#antifa which we traced back to various tweets ei-
ther blaming or defending the Antifa movement. One
hashtag theme in the El Paso only covered the de-
bate about white supremacy and terror attacks car-
ried out by white citizens (#whitesupremacistterror-
ism, #whitenationalistterrorism).
For Santa Fe, we identified one irrelevant hash-
tag cluster, consisting of hashtags about the royal
wedding, Syria and racism. Furthermore, we found
gun control themes mixed with general information
(e.g., #santafestrong,#santafehighschool”). Another
more defined cluster arose with #nrabloodmoney and
#neveragain, mainly blaming the NRA. Interestingly,
many pro-democrat tweets use hashtags usually as-
sociated with the Republican party, e.g. #maga for
“make America great again” or #gop for “grand old
party” but an investigation of sample tweets using
those hashtags found an appeal for voting against Re-
publicans. This practice is also known as “hashtag
hijacking”(Hadgu et al., 2013).
For the shooting at the Youtube Headquarters in
San Bruno we identified one hashtag theme about
the female shooter (light blue cluster in Figure 5d)
as well as clusters about gun control, Donald Trump,
and event-specific entities. This network was the only
one with an alternative hashtag among the most pop-
ular (#qanon) which was found in the same theme as
#kag (“keep America great”, which was used to sup-
port Donald Trump).
6 DISCUSSION
Alternative Narratives Use Nouns with Little Con-
nection to the Event Itself. We found alternative
narratives to be defined by nouns and combinations
thereof that exhibit no direct connection to the event
itself (e.g., George Soros, Israel, George Webb). This
also included news outlets known to spread misin-
formation (e.g., Crowdsource the Truth, #csthetruth,
Jason Goodman). In addition, we found vocabulary
from conspiracy theories, such as Qanon or False
Flag
6
accusations across all data-sets. Based on these
6
False Flag accusations assume that an attack was
planned by an official institution, usually the Government,
to pursue some hidden agenda.
findings, alternative narrative structures can prevail
over multiple events.
Alternative Narrative Patterns Persist Across
Events. We found alternative narratives, such as false
flag accusations, across all data-sets. Additionally,
we saw a strong partisan divide, which was indicated
by hashtags like #maga (make america great again),
or #bluewave (Democrats expecting an election win).
The structural patterns of alternative narratives did not
differ from mainstream narratives, though. Neverthe-
less, patterns reappeared across events, even in the
time-span of two years between the events studied in
this paper.
Alternative Narratives Diffuse no Differently than
Mainstream Narratives. In our user analysis, we
found no obvious difference in message diffusion be-
havior of alternative vs. mainstream messages. On av-
erage, mainstream messages received more retweets.
Such retweets often originate from a few breaking
news messages. We could not identify more favourite
tags for alternative tweets either.
Limitations. Our study comes with several limita-
tions. Firstly, users might try to disturb (”troll”) al-
ternative narratives to upset other users engaging in
them. Klein et al. (Klein et al., 2018) suggest be-
tween 4% - 12% of all users in a Reddit conspiracy
forum just want to annoy or provoke users engaging
in the respective discourse. This should be kept in
mind when working with alternative narratives. Fur-
thermore, our data-sets only cover four gun shootings
in the U.S. which might be too little to draw a conclu-
sion over time and across different events.
7 CONCLUSION
In our study, we analyzed tweets during and in the
immediate aftermath of four gun shootings in the
U.S. between 2017 and 2019. By analyzing main-
stream news articles, we identified mainstream nar-
ratives which were prevalent in the wake of each
event. We labeled tweets based on their hashtags ei-
ther as an alternative narrative or a mainstream narra-
tive and carried out a semantic analysis on the text
of the tweet. Thereby, we found alternative narra-
tives include much more unrelated topics without an
obvious connection to the event itself, as compared
to mainstream narratives. We also identified reoccur-
ring themes connected to conspiracy theories, such as
Qanon.
COMPLEXIS 2023 - 8th International Conference on Complexity, Future Information Systems and Risk
24
texas
elpasostrong
hooting
pasoterroristattack
domesticterrorism
guncontrolnow
chicago
daytonstrong
daytonshooting
dayton
whitenationalistterrorism
whitesupremacistterrorism
prayforelpa
nra
antifa
gunreformnow
walmartshooting
trum
massshootings
(a) Hashtag Co-occurrence network for El Paso.
ol
rvegas
step
g
man
pray
pr
(b) Hashtag Co-occurrence network for Las Vegas.
parkland
guncontrolnow
control
xas
nra
santafehighschool
syria
acism
themout2018
nrabloodmoney
gop
nora
royalwed
gunreformnow
neveragain
schoolshooting
maga
santafestron
enoughisen
texasshooting
(c) Hashtag Co-occurrence network for Santa Fe.
iran
i ghdam
shooter
eaking
kag
news
ravelban
t
t
nasimeabz
msm
wednesdaywisdom
alyssam
ma
guncontro
qanon
nra
2a
nasi
breakingne
peta
(d) Hashtag Co-occurrence network for San Bruno.
Figure 5: Hashtag Co-occurrence networks for each event. Nodes in the same color belong to the same thematic cluster.
REFERENCES
Cunliffe, A. L., Luhman, J. T., and Boje, D. M. (2004). Nar-
rative temporality: Implications for organizational re-
search. Organization Studies, 25(2):261–286.
Eriksson, M. (2016). Managing collective trauma on so-
cial media: The role of Twitter after the 2011 Norway
attacks. Media, Culture & Society, 38(3):365–380.
Fisher, W. R. (1984). Narration as a human communication
paradigm: The case of public moral argument. Com-
munication Monographs, 51(1):1–22.
Fong, A., Roozenbeek, J., Goldwert, D., Rathje, S., and
van der Linden, S. (2021). The language of conspir-
acy: A psychological analysis of speech used by con-
spiracy theorists and their followers on twitter. Group
Processes & Intergroup Relations, 24(4):606–623.
Goodwin, R., Lemola, S., and Ben-Ezra, M. (2018). Media
use and insomnia after terror attacks in France. Jour-
nal of Psychiatric Research, 98:47–50.
Grobelscheg, L., Ku
ˇ
sen, E., and Strembeck, M. (2022). Au-
tomated narratives: On the influence of bots in narra-
tives during the 2020 vienna terror attack. In Proceed-
ings of the 7th International Conference on Complex-
ity, Future Information Systems and Risk - Volume 1:
COMPLEXIS,, pages 15–25. INSTICC, SciTePress.
Hadgu, A. T., Garimella, K., and Weber, I. (2013). Political
hashtag hijacking in the u.s. In Proceedings of the
22nd International Conference on World Wide Web,
WWW ’13 Companion, page 55–56, New York, NY,
USA. Association for Computing Machinery.
Hardy, M. M. and Miller, B. M. (2022). Memorializ-
ing Tragedy on Twitter: Analyzing #PRAYFOROR-
LANDO Following the 2016 Pulse Night Club Shoot-
ing. Communication Studies, 73(2):136–150.
Klein, C., Clutton, P., and Dunn, A. G. (2019). Pathways to
conspiracy: The social and linguistic precursors of in-
volvement in reddit’s conspiracy theory forum. PLOS
ONE, 14(11):1–23.
Klein, C., Clutton, P., and Polito, V. (2018). Topic modeling
reveals distinct interests within an online conspiracy
forum. Frontiers in Psychology, 9.
Ku
ˇ
sen, E. and Strembeck, M. (2021). Building blocks of
communication networks in times of crises: Emotion-
exchange motifs. Computers in Human Behavior,
123.
Ku
ˇ
sen, E. and Strembeck, M. (2021). Emotional Communi-
cation During Crisis Events: Mining Structural OSN
Patterns. IEEE Internet Computing, 25(02):58–65.
Li, J., Tang, J., Liu, X., and Ma, L. (2019). How do users
adopt health information from social media? the nar-
Mainstream and Alternative Narratives in the Wake of Gun Shootings
25
rative paradigm perspective. Health Information Man-
agement Journal, 48(3):116–126. PMID: 30246551.
Nied, A. C., Stewart, L., Spiro, E., and Starbird, K. (2017).
Alternative narratives of crisis events: Communities
and social botnets engaged on social media. In Com-
panion of the 2017 ACM Conference on Computer
Supported Cooperative Work and Social Computing,
CSCW ’17 Companion, page 263–266, New York,
NY, USA. Association for Computing Machinery.
Radzikowski, J., Stefanidis, A., Jacobsen, K. H., Croitoru,
A., Crooks, A., and Delamater, P. L. (2016). The
measles vaccination narrative in twitter: A quantita-
tive analysis. JMIR Public Health Surveill, 2(1):e1.
Shearer, E. and Mitchell, A. (2021). Broad agreement in
u.s. even among partisans on which news outlets
are part of the ‘mainstream media’. Technical report,
Pew Research Center.
Starbird, K. (2017a). Examining the alternative media
ecosystem through the production of alternative narra-
tives of mass shooting events on twitter. Proceedings
of the International AAAI Conference on Web and So-
cial Media, 11(1):230–239.
Starbird, K. (2017b). Examining the alternative media
ecosystem through the production of alternative narra-
tives of mass shooting events on twitter. Proceedings
of the International AAAI Conference on Web and So-
cial Media, 11(1):230–239.
Stieglitz, S., Bunker, D., Mirbabaie, M., and Ehnis, C.
(2018). Sense-making in social media during extreme
events. Journal of Contingencies and Crisis Manage-
ment, 26(1):4–15.
Straka, M. and Strakov
´
a, J. (2017). Tokenizing, POS tag-
ging, lemmatizing and parsing UD 2.0 with UDPipe.
In Proceedings of the CoNLL 2017 Shared Task: Mul-
tilingual Parsing from Raw Text to Universal Depen-
dencies, pages 88–99, Vancouver, Canada. Associa-
tion for Computational Linguistics.
Tausczik, Y. R. and Pennebaker, J. W. (2010). The psy-
chological meaning of words: Liwc and computerized
text analysis methods.
van Prooijen, J.-W. and van Dijk, E. (2014). When conse-
quence size predicts belief in conspiracy theories: The
moderating role of perspective taking. Journal of Ex-
perimental Social Psychology, 55:63–73.
Wang, Y., Han, R., and Lehman, T. (2022). Do Twitter users
change their behavior after exposure to misinforma-
tion? An in-depth analysis. Soc. Netw. Anal. Min.,
12(167).
Weick, K. E. (1995). Sensemaking in organizations, vol-
ume 3. Sage.
Zappavigna, M. (2015). Searchable talk: the linguistic func-
tions of hashtags. Social Semiotics, 25(3):274–291.
COMPLEXIS 2023 - 8th International Conference on Complexity, Future Information Systems and Risk
26