Do not forget to share this picture...the social lives of solidarity hashtags

Team Members

Marloes Geboers (Visual Methodologies, Amsterdam University of Applied Sciences)

Jelmer Kos (Journalism MA, University of Amsterdam)

Rick Plantinga (Journalism MA, University of Amsterdam)

Carlos d’Andréa (Federal University of Minas Gerais, Brazil)

Maja Sawicka (University of Warsaw)

Shefali Bharati (New Media and Digital Culture, University of Amsterdam)

Sandra Dupuy (Communication for Development, Malmö University)

Constance Duncombe (The University of Queensland, Australia)

Serena Del Nero (DensityDesign, Politecnico di Milano)

Contents

Summary of Key Findings

When we are looking at the social lives of hashtags over time, we can see that hashtags that call out for action (#IranOutofSyria, #PrayforSyria and #resist) get resurrected whenever a disastrous event on the ground takes place (in our research this was the Khan Sheikhoun chemical attack in Syria of April 2017). Especially PrayforSyria took up when this tragedy took place, from being almost dead to being the largest of the three solidarity hashtags). From looking at the images that are tied to this particular hashtag we can see how symbolic, more generic, images (black ribbon, hearts and text based calls for solidarity images) are on the foreground at the peak of user activity, during and right after the attacks took place.

We also analyzed the images that were tweeted in the days (12 days) after the attacks took place and see how two, much more specific and at times explicit categories of images take over: infant victims and images with a political message. Note, however, that the time span of 12 days is quite limited and it would be of great interest to research (re)tweeted images relating to this solidarity hashtag over a longer time span in the aftermath of this tragedy.

Interesting visual categories we encountered were collages of children that depict a before and after situation (the children being alive and dead respectively). This is a genre that is quite different from what is known in humanitarian campaigning. Twitter being a platform following what's happening and thus playing a big role in the news sphere, one should also take into account research into the impact of news images, namely Barbie Zelizer's work (2010). Zelizer describes a visual strategy within journalism of depicting images of people 'about to die', to maximize public impact. However, what users do in user re-appropriations on Twitter is depicting two points in time: way before misery set in and the actual death itself. The children images, especially in the aftermath of the event, were picked up quite extensively and were thus being retweeted intensely (see tree map).

When we look at the emotions that are conveyed, we see how sad and fear are structural emotions and during the event other emotions occur (love particularly). In the images a recurrent combination of emotions is conveyed and that is love and sadness (which is the second most correlated emotion found in the Facebook Reactions project of the summer school of 2017).

An interesting question arises from this exploratory project:

In 'flat' periods of user activity, how do people on the ground, in the place of actual suffering, use the solidarity hashtag to still attract attention for their cause? See image below, which was, sadly enough only tweeted or retweeted six times..

image_6ocurrencies_chunk1_prayforsyria_dontforgettoshare.jpg

1. Introduction

As a way of following up on emotion research into Facebook (https://wiki.digitalmethods.net/Dmi/EmotionalClicktivismTheSequel) which was undertaken in a DMI summer school of 2017, this project engages into exploring ways of doing (visual) emotion research on Twitter. By using and following hashtags and their connecting tweet content relating to the Syrian conflict, we explored the visual language of emotionally laden hashtags in the Twitter space and their user re-contextualizations. This way we gain insight in how different tag communities use image content to convey emotional expressions.

In Affective Publics (2015), Zizi Papacharissi, sets out a theoretical model for understanding affective publics: public formations that are textually rendered into being through emotive expressions that spread virally through networked crowds. Affective publics are defined as networked public formations that are mobilized and connected or disconnected through expressions of sentiment. Papacharissi’s research focuses on Twitter text, but communication these days on social media is increasingly undertaken through images. Just as using social buttons such as Facebook Reactions and the heart shaped Like button in Instagram, the way users produce, recontextualize and distribute images to engage with societal issues is part of contemporary connective action (Bennett & Segerberg, 2012).

Recontextualizing images themselves by modifying the visual content and/or recontextualizing visuals through the use of text is a modality of user engagement that cannot be ignored. This project focusses on the user recontextualizations of image material that is connected to three hashtags in the Twitter space: #PrayforSyria, #IranoutofSyria and #resist. The selection of these hashtags and the study of their user recontextualizations are elaborated upon in the sections below.

2. Initial Data Sets

We started off with one of the largest TCAT datasets available in DMI's servers: the bin with the keyword "Syria" with 133,566,442 collected tweets since November 23, 2011.

This was subsetted to a time span surrounding the chemical attacks on Khan Sheikhoun in April 2017 (Feb 1 - Sep 1 2017) giving us 21.253.701 tweets.

Based on a selection of three hashtags that were 'calling for action' (see selection procedure below) we ran three queries that gave us three datasets that could be used for the image-hashtag networks that show the visual language connected to these tags.

In a following stage (qualitative) we narrowed down our analysis to the hashtag that significantly took up in the event time span (second time span, see below) and that was #PrayforSyria.

3. Research Questions

What are the visual characteristics of images tied to hashtags concerning Syria?

Can we detect visual patterns connected to certain hashtags?
Do these images convey particular emotions?

What role do user re-appropriations play in the social sharing of emotions on Twitter?

Can we derive from ‘emotional images’ and their retweet data, shared attitudes towards the Syrian crisis in hashtag communities?

4. Methodology

1) Based on a 21.253.701 tweets' subdataset collected between Feb 1 - Sep 1 2017, we identified on TCAT the "Top Hashtags" related to the keyword "Syria" during 3 periods of time:

Before the chemical attack - March 23 to April 3
“During” the chemical attack- April 4 to April 7
After the chemical attack - April 8 to April 20

Among the most popular hashtags, some were identified as "call for action" tags, such as #resist, #IranOutofSyria and #PrayforSyria.
We focused on the latest, which appeared on all periods of time.

2) We identified the most shared images ("Media Frequency" on TCAT) associated with #PrayforSyria in each period of time.

These images were plotted in a Visual TreeMap (image X) and in an Image-Hashtag network (image X).

180111_treemap.jpg

180111_network-chunk3.jpg

These protocol summarizs these metodological procedures:

treemaprotol.jpgimage network_protocol.png

To identify the emotional content of images, the project's team classified feelings associated with the most popular images (image X)

emotional classification many images.png

3) We decided them to take a closer look at one image that evokes emotion: a cartoon that describes the sad history of a refugee (Image XX).

cartoon.png

Due to a new functionality implemented on TCAT by Emile den Tex, we could then query the subdataset to identify the three tweets in which this image was embedded (IMAGE XXX) and trackback how two of them were retweeted over time (Video X).

3 tweets cartoon.png

5. Findings

see presentation slides

@all: adding the visuals is explained by screenshots in a sub folder on the drive (a how to folder)

I (MG) can work on this as of January 23 ...

6. Discussion

There are several points of discussion and some limitations.

As we have no arabic tags or key words in our query bin on TCAT, the date is most likely skewed towards the western perspective. It is of great interest to start a query bin that includes relevant Arabic hastags and keywords.

Also, in the user network analysis, on the surface level we see that the emotionally laden retweet (includes an emoji and more emotional language) might be retweeted more intensely, however from a close look into the user stats it is also possible that the less emotional frame was still shared more (retweet/follower ratio).

> fake followers
> offline factors of influence potentially

@Constance: could you elaborate on the abovementioned issue a bit here? THANKS!!!!

7. Conclusions

When we are looking at the social lives of hashtags over time, we can see that hashtags that call out for action (#IranOutofSyria, #PrayforSyria and #resist) get resurrected whenever a disastrous event on the ground takes place (in our research this was the Khan Sheikhoun chemical attack in Syria of April 2017). Especially PrayforSyria took up when this tragedy took place, from being almost dead to being the largest of the three solidarity hashtags). From looking at the images that are tied to this particular hashtag we can see how symbolic, more generic, images (black ribbon, hearts and text based calls for solidarity images) are on the foreground at the peak of user activity, during and right after the attacks took place.

We also analyzed the images that were tweeted in the days (12 days) after the attacks took place and see how two, much more specific and at times explicit categories of images take over: infant victims and images with a political message. Note, however, that the time span of 12 days is quite limited and it would be of great interest to research (re)tweeted images relating to this solidarity hashtag over a longer time span in the aftermath of this tragedy.

Interesting visual categories we encountered were collages of children that depict a before and after situation (the children being alive and dead respectively). This is a genre that is quite different from what is known in humanitarian campaigning. Twitter being a platform following what's happening and thus playing a big role in the news sphere, one should also take into account research into the impact of news images, namely Barbie Zelizer's work (2010). Zelizer describes a visual strategy within journalism of depicting images of people 'about to die', to maximize public impact. However, what users do in user re-appropriations on Twitter is depicting two points in time: way before misery set in and the actual death itself. The children images, especially in the aftermath of the event, were picked up quite extensively and were thus being retweeted intensely (see tree map).

An interesting question arises from this exploratory project:

In 'flat' periods of user activity, how do people on the ground, in the place of actual suffering, use the solidarity hashtag to still attract attention for their cause? See image below, which was, sadly enough only tweeted or retweeted six times..

Dominant emotions

When we look at the emotions that are conveyed, we see how sad and fear are structural emotions and during the event other emotions occur (love particularly). In the images a recurrent combination of emotions is conveyed and that is love and sadness (which is the second most correlated emotion found in the Facebook Reactions project of the summer school of 2017).

8. References

List your references in a standard academic bibliographic format.

PRESENTATION SLIDES JANUARY 12 2018

-- MarloesGeboers - 10 Jan 2018
Topic attachments
I Attachment Action Size Date Who Comment
180111_network-chunk3.jpgjpg 180111_network-chunk3.jpg manage 3 MB 15 Jan 2018 - 10:48 CarlosDandrea  
180111_treemap.jpgjpg 180111_treemap.jpg manage 3 MB 15 Jan 2018 - 10:47 CarlosDandrea  
3 tweets cartoon.pngpng 3 tweets cartoon.png manage 136 K 15 Jan 2018 - 10:55 CarlosDandrea  
cartoon.pngpng cartoon.png manage 374 K 15 Jan 2018 - 10:54 CarlosDandrea  
emotional classification many images.pngpng emotional classification many images.png manage 59 K 15 Jan 2018 - 10:52 CarlosDandrea  
image network_protocol.pngpng image network_protocol.png manage 27 K 15 Jan 2018 - 10:39 CarlosDandrea  
image_6ocurrencies_chunk1_prayforsyria_dontforgettoshare.jpgjpg image_6ocurrencies_chunk1_prayforsyria_dontforgettoshare.jpg manage 27 K 11 Jan 2018 - 12:48 MarloesGeboers  
treemaprotol.jpgjpg treemaprotol.jpg manage 17 K 15 Jan 2018 - 10:42 CarlosDandrea  
video-dynamic-net-retweet.mp4mp4 video-dynamic-net-retweet.mp4 manage 1019 K 15 Jan 2018 - 10:58 CarlosDandrea  
Topic revision: r6 - 15 Jan 2018, CarlosDandrea
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