UK Politics and Internet Memes an Analysis of Theresa May Memes Shared on Twitter During the 2017 General Election

Information School

INF6000 Dissertation COVER SHEET (TURNITIN)

Registration Number 160129287

Family Name Gruia First Name Mihaela

Use of unfair means. It is the student's responsibility to ensure no aspect of their work is plagiarised or the result of other unfair means. The University’s and Information School’s advice on unfair means can be found in your Student Handbook, available via http://www.sheffield.ac.uk/is/current.

Assessment Word Count 14,995.

If your dissertation has a word count that is outside the range 10,000 – 15,000 words or if you do not state the word count then a deduction of 3 marks will be applied

Late submission. A dissertation submitted after 10am on the stated submission date will result in a deduction of 5% of the mark awarded for each working day after the submission date/time up to a maximum of 5 working days, where ‘working day’ includes Monday to Friday (excluding public holidays) and runs from 10am to 10am. A dissertation submitted after the maximum period will receive zero marks.

Ethics documentation should be included in the Appendix if your dissertation has been judged to be Low Risk or High Risk. ✓ (Please tick the box if you have included the documentation)

A deduction of 3 marks will be applied for a dissertation if the required ethics documentation is not included in the appendix; and the same deduction will be applied if your research data has not been available for inspection when required.

The deduction procedures are detailed in the INF6000 Module Outline and Dissertation Handbook.

UK Politics and Internet Memes An Analysis of Theresa May Memes Shared on Twitter during the 2017 General Election

A study submitted in partial fulfilment of the requirements for the degree of MSc Data Science

THE UNIVERSITY OF SHEFFIELD

MIHAELA GRUIA

4 September 2017

ACKNOWLEDGMENTS

One spring afternoon, on the 23rd of March 2016, I met Dr Farida Vis in the Diamond Cafe. I was fascinated by her work and our conversation opened my mind to the prospect of completing the Data Science MSc in the Information School. It was a serendipitous encounter that helped kickstart a wonderful journey.

Nearly 3 months later, on the 29th of June, I received a letter from Catherine McKeown from the UoS Finance Department that I had been awarded a Postgraduate Scholarship. Completing this course would have not been possible without that fortunate encounter, and without the scholarship. Thank you from the bottom of my heart.

Although it often left me frustrated, tired, and confused, this dissertation has also tested my new ninja data science skills, and allowed me to use old knowledge that I gained in my Politics & IR BA. I can honestly say that this been my favourite project to work on in my entire academic career.

Thank you Farida for the all the office hours and ‘walk-and-talks’ to the train station, as opportunities to reflect and refine my ideas. I have always admired you, your energy, your ability to multitask on top of multitasking, and coming out at the end of it being the superwoman that you are. The photos of Miss Finn have been the delight of my days when I needed that small push to keep coding.

Thank you Peter for agreeing to be my second supervisor during the crucial last month of August and for reading my work and refining it in the process.

Hannah and Chris, you are the best second coders any social media researcher could ever ask for, thanks for the attention to detail and patience to go through my spreadsheets and code frames. I owe you one so make sure you redeem it.

Gemma, thanks for coming in for the ‘dissertation-work-sessions’, we may not have been the most productive, but we kept each other going and supported one another.

In a rather awkward way, thank you to the Prime Minister for providing us with such interesting visual material for study...

Lastly, a special thank you to my better half, Garrett, for listening me talk about memes of Theresa May for hours on end, for engaging with my work and showing interest in all of it, despite it often being the opposite. I feel proud that you now know what a GIF is.

Structured Abstract

Background Theresa May has been the centre of political controversy since she became UK Prime Minister (in July 2016). An especially controversial decision was her calling a snap election in April 2017, for a General Election (GE) to be held in June 2017. Given the increasing use of social media to manifest political views, images play a significant role in the ways in which political views are expressed. Memes and GIFs in particular are a subset of images that increasingly play an important role in how voters comment on politics on platforms like Twitter.

Aim The overall aim of this dissertation is to better understand the ways in which Twitter users utilised memes and GIFs during the 2017 GE to respond and engage with Theresa May’s campaign and performance as a politician and specifically, as a female politician. In particular, three case studies were selected, each corresponding to a key aspect of her campaign during May and June 2017: strong and stable (SAS), weak and wobbly (WAW) and TheresaMayGIFs (TMG). The aim was to understand what was depicted in the memes, who shared the memes and what was the stance of the memes with regard to Theresa May’s campaign.

Methods The dataset consisted of the 300 most popular shared tweets containing images (100 per case study) and they were collected using Pulsar, a commercial analytics tool. Several social media research methods were employed in the analysis of the memes, namely content analysis of image and text, actor type analysis and sentiment analysis of image and text. Specific code frames were developed for each method and second coders were trained to refine the accuracy of each code frame. The Intercoder Reliability scores of the code frames were reported using Scott’s Pi coefficient.

Results First, Theresa May’s policies and manifesto were revealed to be the most referred to issues by posters in two of the three case studies (SAS and WAW), while Theresa May’s TV performance was the most frequent issue in the TMG case study. Second, members of the public were the most active category of Twitter user in two of the three case studies (WAW & TMG), and for the third case study it was political actors (SAS). Lastly, when analysed both in isolation of the text as well as alongside

the text, all case studies revealed that the memes were critical of Theresa May directly or of an aspect of her campaign.

Discussion These findings suggest the following implications. First, memes can have a political orientation, and this has the potential to be negative. Second, these findings suggest that memes are a powerful form of political expression and that it is difficult for parties to control the narrative online. Third, political parties could consider more clever ways and be savvier when it comes to ways in which they deal with Twitter and memes. Lastly, traditional forms of political campaigning and communication, which may imply repeating the same line or slogan over and over again, are not effective when it comes to certain social media avenues, and particularly Twitter and memes.

TABLE OF CONTENTS

Chapter 1. Introduction and Context pp 7-11 Chapter 2. Literature Review pp 12-20 2.1. Politics and Social Media pp 12-15 2.2. Visual Representation of Female Politicians pp 15-16 2.3. Memes 2.3.a. Conceptual Definition of Memes pp 16-17 2.3.b. Memes and Politics pp 17-19 2.3.c. Memes and British Politics pp 19-20 Chapter 3. Methodology pp 21-34 3.1. Case Studies and Visual Motifs pp 21-28 3.1.1. Strong and Stable 3.1.2. Weak and Wobbly 3.1.3. TheresaMayGIFs 3.1.4. Maybot 3.1.5. Laughing Theresa May 3.2. Methods pp 28-31 3.2.1. Method 1 Content Analysis of Image and Text 3.2.2. Method 2: Actor Type Analysis 3.2.3. Method 3: Sentiment Analysis of Image and Text 3.2.4. Note for All Methods: Intercoder Reliability 3.3. Data Collection Strategy pp 32-34 3.3.a. Tools and Steps to Retrieve Data 3.3.b. Search Terms 3.3.c. Sample Size 3.3.d. Data Storage and Security 3.3.e. Ethics Chapter 4. Findings, Analysis and Discussion pp 35-63 4.1. Findings: Strong and Stable pp 35-43 4.1.a. Content of Memes

4.1.b. Posters of Memes 4.1.c. Evaluative Stance of Memes 4.1.d. Strong and Stable: Discussion 4.2. Findings: Weak and Wobbly pp 43-50 4.2.a. Content of Memes 4.2.b. Posters of Memes 4.2.c. Evaluative Stance of Memes 4.2.d. Weak and Wobbly: Discussion 4.3. Findings: TheresaMayGIFs pp 51-61 4.3.a. Content of Memes 4.3.b. Posters of Memes 4.3.c. Evaluative Stance of Memes 4.3.d. Weak and Wobbly: Discussion 4.4. Comparing Case Studies pp 61-64 Chapter 5. Conclusion pp 65-68 5.1. Review of Research Questions and Research Objectives – Summary of main conclusions and findings 5.2. Limitations of the Research 5.3. Suggestions for Future Research References pp 69-74 Appendix 1 pp 75-77 Code frame 1 RQ1 Appendix 2 pp 78-80 Code frame 2 RQ2 Appendix 3 pp 82 Code frame 3 RQ3 Appendix 4 p 82 [Attached document] Ethics Approval Letter Appendix 5 pp 83-84 [Attached document] Access to Dissertation form Appendix 6 pp 85-87 [Attached document] Confirmation of Address form

Chapter 1: Introduction and Context

Despite taking charge of the UK at one of the most turbulent times in recent political history given the Brexit vote, Theresa May faced intense criticism that she did not hold a legitimate mandate to be Prime Minister (PM). This was because she was handed down the leadership of the country from David Cameron without going through the procedures of a normal election. It was in this political climate that on the 18th April 2017 Theresa May called for a snap election, ahead of 20201, in order to secure a mandate and be in a strong position to negotiate Britain’s relationship with the EU. At the time of calling the election, May was seen as the favourite candidate, thought to be able to secure a majority in Parliament, projecting herself as the embodiment of strong leadership (Parry, 2017, p. 124). Alternatively, Labour leader Jeremy Corbyn was in a somewhat awkward political position, with considerable criticism for being weak and ‘unelectable’. Although the campaign was short, May’s undefeatable position changed within the course of eight weeks and her position weakened. The election results (a hung Parliament, which meant no overall majority for the Conservative Party) forced Theresa May to form a coalition with the Democratic Unionist Party in order to secure a majority in Parliament. This result was generally regarded as a failure, which put her in a weaker negotiating position with the EU than before the election.

This dissertation focuses on Theresa May memes shared during the UK General Election (GE), and the memes selected for study were i) created in response to certain political events or developments in her campaign, ii) contained a meme-specific hashtag, and iii) were the most shared pieces of content on Twitter.

Memes are commonly defined as a piece of culture that can be replicated or generated by humans to help evolve society (Dawkins, 1976). In the context of this dissertation, a ‘meme’ represents a remixed digital object that is copied and spread rapidly by Internet users, often with slight variations, and accompanied by the use of a hashtag. Hashtags (using with the ‘#’ symbol) are a feature of Twitter that is ‘used to index keywords or topics on Twitter’, allowing users to join a discussion on a particular topic (Twitter Support, n.d.). Memes can take the form of static (i.e.

1 This is when the next election was planned for after the 2014 GE. Elections in the UK happen roughly every 5 years, anything in between that is called a ‘snap election’.

images, image macros2) or animated images (i.e. GIFs3). This dissertation adopts the view that what constitutes a meme is the fact that they are widely shared, hence all the attachments (i.e. images, image macros, GIFs, or videos) of the top 100 most retweeted tweets in each case study were considered ‘memes’ and included in the analysis.

The three selected case studies were: 1. ‘Strong and Stable’ (SAS) – linked to the Conservative Party Manifesto and campaign slogan (#strongandstable) This hashtag and the phrase have been consistently used on Twitter from the beginning of May until the election day in June, as it formed part of the campaign slogan ‘Strong and stable leadership in the national interest’ (Conservative Manifesto). The Conservative Party manifesto, launched on the 18th of May (https://www.conservatives.com/manifesto), was also linked to the #strongandstable hashtag on Twitter (Odell & Mance, 2017). In a desire to ‘stay on message’, Theresa May excessively repeated the phrase, thus garnering the reputation of a somewhat robotic character. This hashtag was included in a number of visual representations of the PM, some using the phrases in support of the leader (by official Party and campaign sources, and voters) and some expressing cynicism towards her campaign.

2. ‘Weak and Wobbly’ (WAW) – linked to the manifesto U-turn on social care policy, otherwise known as the ‘Dementia Tax’ (#weakandwobbly) The phrase first appeared on social media after journalist Michael Crick uttered it during a press conference on the 22nd of May in relation to the social care Conservative policy. Shortly after that moment, the phrase was turned into a hashtag, #weakandwobbly, and consistently used by Twitter users that felt disenfranchised with May's campaign and policies. This hashtag, as well as #strongandstable, were included by users in numerous visual representations of the PM, either by superimposing text on original images, or Photoshopping images in different settings.

This online behaviour contributed to turning the pieces of content into viral political memes, nested under the two single and consistent hashtags. This continuous repetition of content highlighted the ‘memetic’ nature of social media, turning visual representations of Theresa May into widely-shared (by virtue of retweets, comments, likes and shares) and recognisable memes.

2 An image macro is ‘a picture with text superimposed on it, that conforms to a template and types of meaning’ (Rintel, 2013:p. 257) 3 The Graphics Interchange Format, i.e. the GIF began as a low-space data format. Today ‘GIF’ is typically used to ‘mean an animated GIF file or an otherwise short, silent, looping, untitled moving image’ (Eppink, 2014, p.298)

3. ‘TheresaMayGIFs’ (TMG) – linked to the live ITV News appearance on the ‘Battle for Number 10’ show’ and the interview with Jeremy Paxman on the 29th of May (#TheresaMayGIFs) This hashtag had the narrowest timespan of all three case studies, because its use on Twitter was almost exclusively used to comment on May's performance on the TV shows mentioned above. The lifespan of the hashtag was roughly 48 hours, and the visual representations of May under this hashtag were predominantly, as the name suggests, GIFs. The hashtag ceased to be used 2-3 days after the TV show, but revealed interesting insights about participatory aspects of social media.

The overall aim of this dissertation is to better understand the ways in which Twitter users utilised memes during the 2017 GE to respond and engage with Theresa May’s campaign and performance as a politician and specifically, as a female politician. In particular, three case studies were selected, each corresponding to a key aspect of her campaign. These were selected in order to better understand how the visual responses created by Twitter users evolved during the campaign and how these memes acted as a form of political, social and gender commentary on her performance as a political leader.

This dissertation specifically focuses on memes shared on Twitter between May 1st and June 9th, a period that covers just over a month before the election day on the 8th of June. In total 300 (100 per case study), most shared memes - in terms of retweets - were individually analysed using content, actor type, and sentiment analyses.

A key aspect of meme-sharing online is the intertextuality that occurs between memes, as certain references can spill over and mix with other key phrases/hashtags creating multi-layered visual interpretations. This was in the case in this analysis of Theresa May memes, in which two additional visual motifs were identified and used as a way to further interpret the case studies and observe the ways in which memes spread online: ● ‘Maybot’ – linked to John Crace’s political sketch published on November 8th 2016, but widely picked up on after the announcement of the snap election in April 2017. As the name indicates, this visual motif is underpinned by the characterisation of Theresa May as a robotic character, because throughout the campaign she excessively repeated the ‘strong and stable’ phrase, and often as a ‘catch-all’ phrase for her position and policies.

● ‘Laughing Theresa May’ – linked to her awkward laugh during Prime Minister’s Questions on March 8th 2017. This visual representation of Theresa May laughing is heavily used on Twitter by users who wish to characterise her as a villainous character.

Through this mapping and analysis this dissertation answers the following three research questions (RQs): ● What is the content of the memes used on Twitter in the three case studies? ● What types of users share those memes? ● What is the evaluative stance of the memes in relation to May’s political leadership?

These research questions correspond to three research objectives (ROs): RQ1 - Objective 1: Understand the kinds of content depicted in these memes and the subject or object they refer to. In other words, what is the descriptive content of the meme and how is this content being presented as a visual experience. Here, the objective is to better understand the types of memes in use (i.e. such as use of images, images with text, GIFs, doctored photos, videos, etc.) as well as content of those memes (i.e. what party leaders they depict, what issues or themes they refer to, what actions they depict). Unlike Objective 3, the goal here is simply to understand the ways in which memes are visually presented, not the position or political view expressed.

RQ2 - Objective 2: Understand the type of Twitter accounts/users engaged in the sharing of political memes of Theresa May during the GE. This refers to the category of user (i.e. media, organisation, member of the public, etc.) as well as the type of political leaning of the account (i.e. left/right-leaning, neutral, official Labour, official Conservative sources). Therefore, examining links between users and political topics can help reveal issues of public concern that different types of Twitter users care about as well as how they engage with those topics via the use of memes.

RQ3 - Objective 3: Understand the ways in which meme-sharing can represent an act of political positioning for people as an expression of their political stance or view (i.e. in support of or critical of Theresa May, her policies or her party). In other words, the dissertation is interested to investigate how the posters of the memes position themselves in relation to Theresa May: critical of, in support of, or neutral. These are the categories that have emerged in answering this research question.

The dissertation is structured into five chapters. Chapter Two includes a literature review that covers three aspects: the theoretical underpinnings between politics and social media; the visual representation of female politicians; and the conceptual definition of a ‘meme’, from its roots in biology to the conceptualisation of the ‘Internet meme’ and its practice on the Internet, particularly in politics. Chapter Three details the methodology of this dissertation. The data for the three case studies was collected from Twitter using Pulsar, a commercial analytics tool. Relevant hashtags and keywords were used. Chapter Four examines the data using multiple methods: content analysis of image and text, actor type analysis and sentiment analysis of image and text, with the aim of understanding how memes are used as a device for political expression and to reveal the kinds of performance evaluations of Theresa May that are encoded in these memes. Where possible data visualisations are used to present the data visually in order to enable the identification of patterns and insights. It is in this chapter that the findings are linked to broader theories identified in the literature review, in order to better understand how memes act as a form of political expression. Lastly, Chapter Five evaluates the findings, discusses limitations, and briefly gives ideas for future research.

The original contribution of this research in the field of social media is twofold. First, this research is timely and speaks to current events, in that it takes place only three months after the 2017 GE. Second, to the best of my knowledge, this research represents the first detailed examination of social media images, especially memes, of Theresa May. As a result, this research contributes to the field of social media research in new and immediate ways, and in particular, adds to the emerging field of social media research on images.

Chapter 2: Literature Review

Given the rise of memes that express political views online, it is important to understand what memes ‘say’ about candidates along with the types of arguments put forward that either support or undermine politicians. The dual nature of memes exists because they can have both positive and negative effects. As Broersma (2015) points out in relation to memes shared in during the 2015 UK GE, referring to then Labour leader Ed Miliband: ‘[Memes can work] both negatively, for example when citizens Photoshop ‘bacon Ed’ in famous film scenes, and positively, as became clear when tweeps started posting pictures of themselves struggling with comfort food, using the hashtag #JeSuisEd’ (p. 23).

The focus of this dissertation is to examine Theresa May’s political performance in her role as PM as well as a candidate in the 2017 GE, and the rich visual reaction and responses that the electorate created on social media regarding three key aspects of her campaign. In interpreting these three political moments, three theoretical lenses were employed. First, theories pertaining to the relationship between politics and social media. Second, the visual representation of female politicians in the UK. Lastly, the realm of memes, images and visual responses when engaging with politics on social media. Together these theoretical lenses can help us to better understand the visual responses shared online in response to Theresa May and her evaluation as a political leader by the Twitter electorate.

2.1. Politics and Social Media

Social media has drastically changed the way campaigns are run and how voters interact with candidates and elected officials. In politics, social media has made politicians more accountable and accessible to voters, and has enabled more nuanced interactions based on rich analytics and real time data at almost no cost. At the same time, social media has been used as a tool to criticise politicians and comment on their policies, actions and statements (Pickard, 2016).

The election that pioneered the use of social media in campaigns was the 2008 US Presidential election between Barack Obama and John McCain. Obama was the first political candidate to effectively use the Internet to organise supporters, advertise and communicate with individuals in a

way that had been impossible in previous elections (Miller, 2008). Following from this, the 2012 US election, the 2015 UK election, and other campaigns around the world, have employed social media and integrated it into their political communications and campaign strategies. For example, Twitter played a key role in the 2013 German election (Isasun, Twiter Blog, 2013), the European Parliament elections in 2014, as well as referendum campaigns across Europe in Spain (2015), Italy (2016) and Greece (2015).

In particular, Twitter has become an effective platform used in political campaigning. There are many features that make Twitter the ‘go to’ platform for political conversation, commentary and breaking news activity, with unparalleled immediacy and speed. A noteworthy concept that explicitly links Twitter and politicians is ‘Twitter diplomacy’ which refers to the use of Twitter by heads of state, diplomats, public officials and governmental organisations (Kelemen, 2012). As a direct extension of this phenomenon, Twiplomacy (twiplomacy.com) is a company that conducts research on the presence of political actors on Twitter and other social media platforms, publishing yearly studies that bring insights into how different social media networks are used by governments. In their 2017 study, Twiplomacy analysed which politicians are on social media and how those individuals are using Twitter, with a focus on the use of Twitter by governments. The study highlights that Twitter is the social media platform of choice for governments and foreign ministries, with ‘856 Twitter accounts belonging to heads of state and government, and foreign ministers in 178 countries, representing 92 percent of all UN member states, with a combined audience of 356 million followers.’ (Twiplomacy Study, 2017, online source). This indicates that officials regard Twitter as an effective tool for communication and representation.

According to Jensen (2017), there are three kinds of empowering communication possible within Twitter’s platform. First, campaigns ‘may engage in dialogue with others. […] Second, the campaigns may retransmit the communication produced by supporters. […] Third, they may invite persons to contribute in the campaign on their own terms, with little guidance from the campaign itself’ (p. 29). Using language that is specific to Twitter, these three actions can be described as replying, retweeting and inviting persons to participate freely. In the context of the 2015 GE, there is little evidence to suggest that social media was used to empower campaign supporters (Jensen, 2017). Moreover, although during the 2015 GE, official parties replied to and retweeted content from members of the public, this was only a small part of their overall activity on Twitter (Jensen, 2017). This suggests that political parties use Twitter more as a broadcasting tool and less as an opportunity for active engagement with their supporters or critics.

In contrast with political parties, TV shows and news channels have capitalised on the ability of social media to be used both as a tool for broadcasting content as well as engagement with the public. This is evident through the concept of ‘second-screen’ experiences, a phenomenon that relates to the ability to experience a TV show both on the TV as well as a second device, usually via social media (Giglietto & Selva, 2014; Gorkovenko & Taylor, 2014). Twitter has been particularly effective at forging this link between the TV and the online space, creating collaborations with shows that need to drive ‘tune in’ (Twitter Marketing, 2012). The link between the two experiences is often the invitation to participate with content or comments with a certain hashtag or by sending in requests or information, in essence being a solicitation for engagement and participation. In the UK, this practice has been the subject of the research in the content of the 2015 UK GE, with evidence to suggest that this practice is gaining more and more momentum and that indeed social media might be the ‘first screen’ that people experience political shows on (Gorkovenko & Taylor, 2014; Anstead & O’Loughlin, 2010). This participatory nature of social media and its symbiosis with televised show informs the analysis of one the case studies, TheresaMayGIFs.

Although social media clearly represents a new way of ‘doing politics’, the growth of social media also represents new challenges and opportunities for political parties. The literature covering the scope and benefit of digital media in politics is split into largely two camps. On the one hand, critics view digital media as dramatically limiting the capacity of ordinary citizens to influence wider debates (Hindman, 20009), as well as ‘structured interactivity’ (Kreis, 2011:11) or ‘controlled interactivity’ (Stromer-Galley, 2014:14). Within this view, power remains in the control of the party system, which is able to formally set the agenda and frame policies, strategies and tactics. On the other hand, researchers have also argued for a ‘democratising effect’ of social media on political campaigning, with scholars like Lilleker and Jackson (2010) arguing that social media creates a participatory architecture that facilitates interactions between campaigns and supporters that otherwise do not exist. In addition, it has been argued that social media spaces and Web 2.0. platforms enable the creation of ‘citizen-initiated campaigning’, producing messages that were not directed by the party (Gibson, 2015), and which allow individuals to self-organise and operate in communities of peer production where individual contributions are valued (Benkler, 2016). However, despite some agreement about decentralised messaging and interactive collaborations within modern political campaigns, most scholars continue to argue that epistemic authority largely remains with the campaign (Jensen, 2017).

With respect to the promotion of May’s epistemic claim to lead the party in the 2017 GE, Parry argues that Theresa May projected a ‘one-dimensional public image’ (2017, p. 124) underpinned by a rather inflexible leadership style. The use of social media in May’s campaign also lacked the humanising, grassroots elements that are characteristic of social media, with one scholar remarking that ‘nobody has a selfie with Theresa May’ (Andrews, 2017). In contrast, Corbyn’s campaign made extensive use of social media to portray him as a three-dimensional leader representing the country, with the politician referring to social media as enabling the ‘accessible, participatory politics for the 21st century’ (Parry, 2017, p. 124). Although a comparison between the two party leaders is beyond the scope of this dissertation, examining the way they used social media to campaign and mobilise the electorate is revealing of their attempts to claim the ‘authority to rule’, and informs the analysis of social media responses later in this dissertation. Overall, Theresa May was seen as missing an opportunity to engage with the public in a participatory way that could shore up her authority to lead, in a way that allowed people to interact with her on a human level – resulting in characterising her as having an inflexible robotic leadership style, which persisted throughout her campaign and underpinned the memetic visual responses that the public produced on social media (Parry, 2017).

2.2. Visual Representation of Female Politicians

The topic of how women are represented in British politics has been subject to extensive research and analysis, from the representation in Parliament of female MPs (Celis and Childs, 2008; Chaney, 2014) to the bias against women in the media coverage of politics (Childs, 2004).

When thinking more specifically about the representations of female politicians in UK elections, three main areas stand out.

First, there has been considerable research dedicated to the first female PM in the UK, Margaret Thatcher. Thatcher was the first female PM and has become the archetype of female politicians against which other prominent female politicians are compared. This was the case in 2015, when First Minister Nicola Sturgeon was compared to Thatcher for the embodiment of both ‘grace’ and ‘steel’ (Dathan, 2015). It is therefore crucial to understand the ‘mark’ left behind by Margaret Thatcher and to understand how she was portrayed as the first woman leader of any UK political party. In understanding the legacy of Thatcher, it is possible to better develop an understanding of how subsequent politicians are compared and relate to Thatcher.

Second, research indicates that the media plays a crucial role in the ‘gendering’ of UK politics, with a heavy emphasis on promoting male-specific qualities as success factors in the political arena. Drawing on studies from Lynch & Dolan (2013) and Campus (2013), masculine traits include being portrayed as competent, decisive and strong, while femininity is associated with compassion, honesty and warmth. In their research, Harmer et al (2016) reveal how the 2015 UK leadership debates were highly gendered and heavily favoured ‘strength and aggression’ as positively associated with masculinity, and therefore an indication of a man’s ability to be a leader (p.14). Particularly, their research reveals three key themes that can provide a useful starting point when interpreting the representation of Theresa May in the memes on Twitter: strength versus weakness, emasculation versus feminisation, deliberation versus combat.

Lastly, research focusing on the ‘performance of gender’ in leadership debates in the 2015 GE reveals interesting insights on how ‘masculinity’ is posited as a criterion for the evaluation of politicians, and how the success of a female politician relates to her successful display of ‘enough’ male qualities for her to be considered a strong leader (Harmer, et.al., 2016). This is especially relevant to Theresa May, given the centrality of her ‘strong and stable’ messaging and the fact that she was viewed as a clear favourite and strong leader when she called the snap election in April 2017.

These three aspects play a key role in the visual representations of Theresa May as a leader and in the discourse used to evaluate her performance as a politician - in the memes, in the text used to share those memes, as well as the hashtags created to share those memes.

2.3. Memes

2.3.a. Conceptual Definition

Memes represent a unique form of social political communication because they are situated at the nexus of language, society, popular culture, communication science and digital technologies (Ross, Rivers, 2017, p.1). The term meme was introduced by the Oxford zoologist Richard Dawkins in his book The Selfish Gene (1976). He called these adapting elements of culture, memes, after the Greek word ‘mimeme’ meaning imitation (Dawkins, 1976). Twenty-three years later, in her book, The Meme Machine, Susan Blackmore (1999) argued that an understanding of memes relies on understanding that memes, like genes, are replicators and the success of a meme depends on its replication or copied behaviour among humans. As Blackmore (1999) argues, ‘everything that you

have learned from imitation from someone is a meme’ (p. 6). Some key examples of memes and replicated behaviour suggested by both Blackmore (1999) and Dawkins (1976) include fashion, music, and other cultural artefacts such as religion or politics. These examples support the idea that memes can be anything within culture that can be quantified and replicated. However, more recently, the term has been adapted to examine a new popular cultural social media related phenomenon.

In 2013, Shifman defined internet memes as ‘units of popular culture that are circulated, imitated, and transformed by individual Internet users, creating a shared cultural experience in the process’ (p. 367). Later on, in her book, Memes in Digital Culture, Shifman (2014) argued that the word meme could be used as a relevant term to understand a wide variety of contemporary behaviours. Knobel and Lankshear (2006) defined memes as ‘widely propagated ideas or phenomena’ online (p. 217). The word meme is now used as shorthand for Internet memes, or images, image macros, GIFs, videos, and email chains online.

2.3.b. Memes and Politics

Memes used in a political context are ‘a site for understanding audiences, media flows and the circulation of popular culture and politics [and they also] act constitutively and work to make salient disparate media narratives and information within a networked culture’ (Burroughs, 2013, p. 259). As is the case for the memes analysed in this dissertation, they have spread throughout the election by conveying messages through a visual-discursive combination of image and text.

The background of Internet memes shared on social media during political elections generally relates to the fact that social media has grown rapidly to the point where it has ‘become one of the most popular Internet services in the world’ (Gil de Zúñiga et al., 2012, p. 319) and thus provides ‘an avenue for social and political participation to many who previously may not have found such an avenue either apparent or available’ (Ross & Rivers, 2017, p. 1). Moreover, there is an emerging body of literature that focuses particularly on the use of memes in a political context. Three studies stand out as useful for this dissertation.

First, Ross & Rivers (2017) selected eight memes on key policy issues for the main presidential candidates of the 2016 US election and examined how memes were frequently used as a device for de-legitimisation. In doing so, they used Van Leeuwen’s (2007) framework for analysing discourses

of legitimation, but employed the framework from an inverted ‘negative’ position – as a form of de- legitimisation. The analysis revealed that the memes shared in the 2016 US election used strategies and techniques to deligimitise and undermine both leaders’ (i.e. Trump and Clinton) credibility. Second, Burroughs (2013) focused on a series of memes depicting President Obama as unpatriotic (i.e. the ‘crotch salute’, the ‘left-hand salute’, and the ‘Veterans Day non-salute’) in order to understand the role of trolling in the public sphere and Internet politics. Through this, Burroughs (2013) tracks ‘the cultural practices and logics of ‘sharing’ political memes and conceptualises memes as part of an agonistic public sphere and media ecology’ (p. 258). He applies a theory of articulation to the study of memes, drawing upon Hall’s (1973) communication theory and the process of circulation for memes. Third, Milner (2013) analysed the use of memes in the political discourse surrounding the Occupy Wall Street movement (OWS) and examined how memes were used to articulate perspectives on OWS. Milner analysed these memes using multimodal critical discourse analysis. His findings indicated that ‘memes facilitated conversation between diverse positions’ and that OWS memes ‘employed populist argument and popular texts, intertwining them into a vibrant polyvocal public discourse’ (2013, p. 2357).

Memes can take different formats. One such format is the GIF. A GIF refers to ‘an animated GIF file or an otherwise short, silent, looping, untitled moving image’ (Eppink, 2014, p.298). GIFs have become a kind of language used in online networks and are used as reactionary tool and commentary, usually in a humorous and ironic way. In this context, ‘reaction GIFs’ specifically capture emotional and bodily reactions and are meant to be inserted into the flow of conversation (Huber, 2015). They are blank canvasses in the sense that they can fit the view, beliefs and perspective of the author using it. For example, a GIF of someone falling can be used as a metaphor for a faux pas of a politician on either left or right of the political spectrum, as long as the text attached to it supports that GIF. Moreover, compared to text and static images, ‘animated GIFs may be especially good at conveying complex emotions because of the greater range of expression of animations and the resemblance to real-life scenarios’ (Jiang and et, 2017, p. 2).

There are several ways in which users can share GIF-containing tweets. The first and easiest way to do so is when they compose a tweet, using a button appropriately titled ‘GIF’ gives users the options to choose from a menu of 44 categories. It is unclear how many GIFs each of these categories contain, because Twitter connects to at least 2 GIF platforms such a Tenor (https://tenor.com/) and Giphy (https://giphy.com/) to obtain content. Users click on categories and select a GIF or they can

simply type words in the search bar available and memes come up that are tagged with those words. The second way is similar to the option of attaching an image to the tweet from a user’s own computer. GIFs are easy to save and they can be stored on computers and attached in the same way as one would an image (Twitter Support Posting a GIf to Twitter, n.d).

2.3.c. Memes and British politics

When it comes to British politics, both social media and memes have made a strong appearance in the 2015 GE, with Twitter being the preferred and most effective form of social media communication and engagement in political campaigns. For example, Twitter’s Head of News, Government and Elections characterised the 2015 GE as the UK’s first ‘Twitter election’ (Charles, 2015, p. 67). Moreover, in the 2015 election, Twitter ‘enabled a multi-vocal (or rather multi-visual) creative expression to emerge, where humorous memes and personalised messages […] featured heavily’ (Parry, 2015, p. 88).

This multi-visual expression focused mainly on the two leaders of UK’s largest parties, David Cameron and Ed Miliband. The visual representations manifested themselves in ways that are characteristic to Twitter, and occurred as a result of certain political events or actions. Social media allowed the electorate to push back against press agendas and create ‘citizen-led campaigns to counter press power through parody and self-effacement’ (Jackson, Thorsen, 2015, p. 8). For example, the 2015 GE was notable for memes such as ‘#Milifandom’, an online campaign in which particularly teenage girls showed support and admiration for the Labour Party leader (Ratcliffe, 2015). This campaign was created as a backlash to the negative portrayals of Miliband in the media, and trended in April 2015. This campaign was picked up by Buzzfeed on the 21st of April 2015 and the next day The Guardian covered the story, a pattern that has also been noticed with the selected Theresa May case studies. Another notable example from 2015 is the ‘#JeSuisEd’ meme which was created in response to Ed Miliband being caught awkwardly eating a bacon sandwich. Users on Twitter shared photos of themselves awkwardly eating a similar sandwich, in an effort to push back on the negative coverage of Miliband in the press. The ‘#JeSuisEd’ moment was juxtaposed with memes created of David Cameron who was photographed eating a hotdog with a knife and fork.

Overall, the use of memes in the 2015 GE, particularly for the Labour Party, were used to demonstrate connection and sympathy, with the ‘Milifandom’ meme being an example of an ‘emergent, contingent and authentic’ meme (Hills, 2015, p. 89, italics in original). Nevertheless, the

meme-sharing did not necessarily translate in vote share and seats. This is often referred to as ‘hashtag activism’ (Khan-Ibarra, 2014) under which people use Twitter hashtags to engage with politics, but fail to cast their vote and make a difference in the real world. What is interesting is how these memes enable or hinder a candidate’s ability to connect with voters and they ways in which voters themselves use memes to promote, criticise or evaluate the performance of politicians.

This literature review has covered the following aspects: theories pertaining to the relationship between politics and social media; the visual representation of female politicians in the UK; and, the realm of memes, images and visual responses when engaging with politics on social media. Together this literature forms the theoretical underpinning of this dissertation and will help address the three research questions regarding the content of the memes; the types of people sharing them and their political leaning; and, the stance conveyed by these users with regards to Theresa May. The methods used to carry out this research are discussed in the next chapter, namely content analysis of image and text, actor type analysis and lastly, sentiment analysis of image and text.

Chapter 3: Research Methodology and Methods

The aim of this chapter is to outline the methodology and methods used in this research. This chapter is structured into three main sections. First, it describes the three case studies and the two visual motifs selected for analysis explaining what the textual references relate to and providing relevant memes as visual examples. Second, it details the methods used in the analysis, namely content analysis of image and text, actor type analysis and lastly, sentiment analysis of image and text, and explains how the code frames were built and tested. Lastly, this section outlines the data collection strategy, in terms of tools and steps undertaken to retrieve the data, the search terms used and the sample size. It also addresses aspects related to data storage and security, as well as the ethical considerations involved in studying social media and the ethics approval obtained for conducting this research.

3.1. Case Studies and Visual Motifs

Conducting Internet-based research on volatile cultural phenomenon such as memes comes with a distinct set of challenges. Using the 2017 UK GE campaign as the parameters for the field of study has helped to narrow and focus the dissertation. Moreover, this dissertation has adopted a case study approach. Below follows a detailed description of the three selected case studies.

KnowYourMeme (http://knowyourmeme.com/) was used as a source of inspiration in writing up the description of the case studies, outlining the origin of the event that sparked the meme, mentioning its spread in the media and showcasing some examples. What is true across all case studies is that the memes became viral on Twitter, were picked up by entertainment and social news publications such as Buzzfeed or CreativeReview, and also picked up in mainstream media outlets such as The Guardian. This spill over from online news into mainstream journalism is a testament to the central role that visual representations, memes and hashtags played on Twitter during May’s campaign.

3.1.1. Strong and Stable

On April 18th 2017, Theresa May came out of 10 Downing Street to announce a snap election. Her speech referred to the need for ‘strong and stable’ leadership in the UK that can help the country navigate the difficult times ahead (BBC, 2017). Throughout the campaign, Theresa May has repeated the strong and stable phrase numerous times (between 18 April - 2 May the PM used the phrase 57 times in speeches (Smith, 2017)) in nearly every interview or to answer reporters’ questions. This repetition managed to gain her a reputation of a somewhat robotic character, which later on contributed to her being characterised as the ‘Maybot’, and resurfacing the political sketch written by John Crace in November 2016. On several occasions during the 2017 GE, Theresa May was accused of being robotic, or of uttering the mantra so many times that it quickly became devoid of any substance.

Although for many the repetition seemed robotic, the communications strategy aligned with what Lynton Crosby, a renowned political strategist involved in numerous Conservative campaigns, would characterise as ‘message discipline’ (Goodman, 2017). Moreover, this campaign strategy for Theresa May was led by election strategist Jim Mesina (involved with the Conservative campaign in 2015 and Barack Obama’s campaigns) who argues that ordinary citizens think about politics for only four minutes a week, and therefore effective political messages needs to be simple, concise and powerful (Bush, 2017).

Throughout the campaign, various images were shared online using the hashtag #strongandstable, or just the phrase ‘strong and stable’ and alluded to the mechanical way in which the PM was repeating the message. A few examples can be found in Images 1a-e.

Images 1a, 1b, 1c

Images 1d, 1e

Sources of Images Image 1a: Jeremy Deller creation, Article by Brown, M. (22 May 2017) Jeremy Deller behind 'strong and stable my arse' posters in London, The Guardian, Available at https://www.theguardian.com/education/2017/may/22/jeremy-deller-strong-and-stable-my-arse-posters- london Image 1b: Shared on Twitter (27 April 2017) Image 1c: Shared on Twitter (27 June 2017) Image 1d: Shared on Twitter (7 June 2017) Image 1e: Gallery by OutrageouslyBritish, 2017 General Memelection Dump (7 June 2017) Imgur, Available at http://imgur.com/gallery/wxTfv

3.1.2. Weak and Wobbly

On the 18th of May 2017, the Conservative Party launched their manifesto in which they outlined their key policies. One of their flagship policies referred to social care and the elderly, stating that that the £72,000 cap on care costs would be scrapped - this cap also became known as the ‘dementia tax’ (Conservative Manifesto, p.65; England, 2017). However, only four days later after the manifesto launch, on May 22nd, it became clear that the Conservatives were less firm about scrapping the cap and were considering its inclusion as an option that would be discussed after the GE (Simons, 2017). This political move was referred to by the press as the ‘dementia tax U-turn’ and was first characterised as ‘weak and wobbly’ by Channel 4 journalist Michael Crick during a press

conference on the 22nd of May 2017 (RT UK, 2017). Image 2a is a screenshot from his Twitter account.

(Image 2a - screenshot from Michael’s Crick’s tweet, 22 May 2017, https://twitter.com/michaellcrick/status/866604346351509504)

From that moment, the phrase was picked up by the press and the Labour party as an antithetic phrase to her ‘strong and stable’ mantra (Hines, 2017; BBC, 2017). Shortly after that, images were circulated online that depicted illustrations in connection to the ‘weak and wobbly’ phrase. Mainstream popular culture and news publications such as The Daily Beast ran stories about the transition from ‘strong and stable’ to ‘weak and wobbly’ for Theresa May. A few examples can be found in Images 2b-e.

Images 2b, 2c

Images 2d, 2e

Sources of Images Image 2b: Shared on Twitter (8 June 2017) Image 2c: Me.Me website https://me.me/i/weak-w-o-b-b-l-y-turn-w-o-14279375 Image 2d: Shared on Twitter (11 June 2017) Image 2e: Shared on Twitter (6 June 2017)

3.1.3. TheresaMayGIFs

On the 29th of May 2017, Theresa May and Jeremy Corbyn participated in the ‘Battle for No 10’ TV show, in which they were questioned by a live studio audience on Sky News and Channel 4 TV, and then interviewed separately by broadcaster and journalist Jeremy Paxman. This was an opportunity for the leaders of UK’s largest political parties to defend and explain their policies and programmes for Britain’s future (The Guardian Live Feed, 2017). The leaders were interviewed separately, as previously Theresa May had declared that she would not participate in any televised debates. Her argument was that ‘debates where the politicians are squabbling amongst themselves don’t do anything for the process of electioneering’ (Merrick, 2017; Elgot, Martinson, 2017).

During the interview of Theresa May, the public took to Twitter to create, share and distribute memes of the PM in different formats, but predominantly in the form of GIFs (i.e. very short moving

images), and using a sarcastic and critical tone towards her answers and performance, as well as towards Conservative supporters and colleagues. The GIFs often represented 2-3 second clips from popular movies, TV shows or cartoons and formulate amusing, sarcastic or ironic responses and reactions (Harris, 2017). The general sentiment expressed in those GIFs was that overall Corbyn had performed better than May.

There is no way of differentiating the way in which someone attaches a GIF to a tweet. However it can be reasonably assumed that if the GIF exists in the Twitter database, users would have used the platform and not uploaded a GIF from their own computer. The fact that the platform enables users to attach this type of content to their tweets highlights the potential of platforms affordances to shape user behaviour online. Therefore, in engaging with the hashtag #TheresaMayGIFs on the day of the Jeremy Paxman interview, the affordance of the platform made this phenomenon possible and easy. This highlights that the choices made by the platform to allow visual content on the platform by adding visual features has the potential to foster a different mode of visual communication and allows users to engage in a different way. Regarding GIFs, the feature was introduced on Feb 17, 2016 (Reddy, Twitter Blog, 2016).

The #TheresaMayGIFs visual responses were reported in Buzzfeed, indicating the mainstream adoption of this cultural phenomenon in the media (White, 2017). Moreover, news outlets such as The Huffington Post (Harris, 2017), The Guardian (Belam, 2017), The Daily Mirror (Oakley, 2017) and The Daily Telegraph (Graham, 2017) included the hashtag #TheresaMayGIFs in the title of their articles, in which they explicitly compiled the most amusing memes and GIFs that were shared during the debate on social media. One such example of a GIF that was used in relation to the hashtag can be found in Images 3a-c.

Images 3a, 3b, 3c, Source https://giphy.com/gifs/drunk-bunk-bed-1Vv3Nt3pjetMc

3.1.4. Maybot

On November 8th 2016, political sketch writer John Crace published a sketch about Theresa May in The Guardian titled ‘Theresa struggles to take back control – from her own Maybot’ in which he caricatured the PM’s robotic answering and continuous deflection of questions (Crace, 2016). As explained in section 3.1.1., throughout the GE campaign, May has repeatedly been criticised for using the ‘strong and stable leadership’ manifesto slogan to justify her position and policies, while avoiding to provide intricate detail. As will be explained in Section 4.4. in Chapter 4, this visual motif has been cross-referenced in many other memes and is a characterisation of May and her communication style and messaging. A few examples of this visual motif can be found in Images 4a- b.

Images 4a, 4b

Sources of Images Image 4a: https://www.pinterest.com/Spiretone1/brexit-british-politics/ Image 4b: Shared on Twitter (10 June 2017)

3.1.5. Laughing Theresa May

On the 8th of March 2017, during a Wednesday session of Prime Minister’s Questions, Jeremy Corbyn posed a question to Theresa May about an alleged ‘deal’ made between the government and Surrey County Council over social care. In response, May displayed a rather unusual laugh that has been picked up on social media, remixed and shared widely, with many mocking her for appearing villainous. The moment became known as ‘Laughing Theresa May’ and gained a page on the famous meme directory ‘Know Your Meme’. The moment in its original form became a powerful visual motif, that has been used in static (as an image) and animated (as a GIF) form, to criticise the PM. The reproduction of the original GIF can be found in Images 5a-c.

(Images 5a, 5b, 5c, Source http://knowyourmeme.com/memes/laughing-theresa-may)

3.2. Methods

The overall aim of this research was to collect the 100 most shared tweets containing images for each case study and to apply a series of research methods to this Twitter data. The section below details the three methods used and the section that follows highlights the data collection strategy. For each case study 100 images were coded, each taken through the same steps, applying the same three methods to each case study. The methods selected in this dissertation are frequently used for doing this type of social media research and combining them allowed for a deeper insight than was possible when using a single method only (Snelson, 2016).

3.2.1. Method 1: Content Analysis of Image and Text

Content analysis (CA) is ‘a research technique for making replicable and valid inferences from texts (or other meaningful matter) to the contexts of their use’ (Krippendorf, 2004 p.18). In other words, CA refers to the ‘analysis of documents […] that seeks to quantify content in terms of predetermined categories’ (Bryman, 2012, p. 289). Lastly, CA can be employed to categorise aspects that are not easy for a machine to categorise and need human input (Thelwall, 2014).

This method was employed both on the text and image of the tweet in order to answer RQ1 ‘What is the content of the memes?’ and meet RO1 to understand the kinds of content depicted in these memes and the subject or object they refer to.

These categories contained in the code frame were inspired by the text of the tweet and content of the memes. The aim of coding information both in the text of the tweet as well as the image was to understand the subject and object the memes refer to. Moreover, the combination allowed for a deeper and richer interpretation of the meme and its meaning. See Appendix 1 for the final code frame.

3.2.2. Method 2: Actor Type Analysis

Actor type analysis is a research technique used to ‘identify different types of users [and] sorting them into what we call “key actor types”’ (Lotan et.al, 2011, p. 3). The actor type code frame was specifically concerned with identifying different types of actors, that is different types of Twitter users, in the dataset.

This method was useful to answer RQ2 ‘What types of users post memes?’ and meet RO2, namely to understand the type of Twitter accounts/users engaged in the sharing of political memes of Theresa May during the GE. This refers to the category of user (i.e. media, organisation, member of the public, etc.) as well as the type of political leaning of the account (i.e. left/right-leaning, neutral). A further distinction was also made between official Party sources and non-officials sources.

The rich interpretation of the 100 selected memes also came from understanding the type of user who is tweeting the content. Developing this actor type code frame (Krippendorf, 2004) drew heavily upon the code frame developed by Procter, Vis & Voss (2013) in analysing the UK summer riots from 2011 on Twitter (see Appendix 2 for the final code frame). Their code frame was built upon the actor type coding developed by Lotan et.al (2011) who looked at different actor types in relation to Twitter use in the Tunisian and Egyptian revolutions. However, Procter, Vis & Voss (2013) expanded on that code frame and introduced a key category of importance to this dissertation ‘members of the public’ defined as ‘Individuals who provide no link to organization or institution. The account appears to be maintained by a private citizen in their personal capacity, highlighting personal information in their bio.’ (Procter Vis & Voss, 2013, p. 214). Through this method, aggregate-level analysis was conducted on the users to better understand the different types of Twitter users in terms of a series of predefined categories; and what that reveals for the practice of meme-sharing during these elections. A further code frame was developed to understand the political leaning of the account (i.e. left/right-leaning, neutral) - (see Appendix 2 for the final code frame). A further distinction was also made between official Party sources and non- officials sources. This was important to better understand where the support or criticism of Theresa May came from.

3.2.3. Method 3: Sentiment Analysis of Image and Text

Sentiment analysis (SA) is ‘the computational treatment of opinion, sentiment, and subjectivity in text’ (Pang, Lee, 2008, p. 6). SA can be performed automatically or manually to detect the sentiment expressed in a piece of content. Given the visual nature of memes and the difficulty of using automated SA for complex visual material, manual SA was used. Moreover, this allowed more flexibility to build a code frame according to the data, and to be more nuanced in the assessment and aptly deal with more challenging aspects of this dataset, such as sarcasm and irony. In order to fully understand the stance of the content, SA was used on both the meme and the accompanying text of the tweet.

This method was employed both on the text and meme of the tweet in order to answer RQ3 ‘What is the stance of the meme?’ and meet RO3, namely to understand the ways in which meme-sharing can represent an act of political positioning for people as an expression of their political stance or view (i.e. in support of or critical of Theresa May, her policies or her party)

To develop the categories for this code frame (see Appendix 3 for the final code frame), Shifman’s (2013) concept of ‘stance’ was explored. Shifman argues that the ‘stance’ in a memetic form can ‘depict the ways in which addressers position themselves in relation to the text, its linguistic codes, the addressees, and other potential speakers’ (p. 367). To use an example of Theresa May captured eating chips (#strongandstable), some users shared that photo imposing text on the image to portray a derogatory position whilst some defended the PM. The proposed categories through which to judge the stance in the memes were ‘Critical’, ‘Supportive’ or ‘Neutral/Unclear’. Given the highly subjective nature of these categories, the code frame was tested on two separate independent coders who were asked to code the same 20 random items. This strengthened the validity and applicability of the code frame: coding for what the researcher thought the stance was, rather than trying to guess what the user had intended, something that can only be found out by asking the users directly.

3.2.4. Note for All Methods: Intercoder Reliability

For each of the three RQs a code frame was developed. To develop the categories for the code frame, a sample of images was observed inductively (Bengtsson, 2016). Each category was labelled with a unique initial code and description of category. Once a code frame was developed, it was

then tested on a second coder (in some cases a third coder as well), by asking them to code a small random sample of the dataset (i.e. 20 items). This step was undertaken in order to determine the intercoder reliability (ICR) of the code frame. Both results were evaluated using ReCal2 (Freelon, 2010). ReCal2 is ‘an online utility that computes intercoder/interrater reliability coefficients for nominal data coded by two coders’ (Freelon, 2010). More specifically, ICR is a term for the extent to which ‘independent coders evaluate a characteristic of a message or artifact and reach the same conclusion’ (Lombard, 2010).

The results of the ReCal2 test were evaluated based on the Scott’s Pi coefficient, a statistic that ‘corrects for the number of categories in the code, and the frequency with which each is used’ (Scott, 1955, p. 323). In terms of the acceptable thresholds, given that Scott’s Pi is a more conservative coefficient, the value of .70 is acceptable in most situations (Hughes and Garrett, 1990; Neuendorf, 2002).

With respect to the ICR, Table 1 below indicates the Scott’s PI value for each code frame, which as explained in Section 3.2.4, must be above 0.70. Each data item from each case study (300 data items in total) was then taken through each of the categories and coded accordingly. A Scott’s value Pi is reported for each of the variables coded in the code frames.

Table 1: ICR results and Scott’s Pi Values

RQ, Method, Code frame Variable Scott’s PI

RQ1 What is the content of the memes? Party Leader 0.759 Method 1 Content Analysis Of Image And Text Issues 0.938 Code frame 1 - see Appendix 1 Type of meme 0.757

RQ2 What types of users share those memes? Category of user 0.824 METHOD 2: Actor Type Analysis Political leaning of user 0.832 Code frame 2- see Appendix 2

RQ3 What is the evaluative stance of the memes? Stance of meme 0.905 / 0.833 METHOD 3: Sentiment Analysis Of Image And Text (Coder 1/Coder 2) Code frame 3- see Appendix 3 Stance of tweet 0.777 / 1 (Code 1 / Coder 2)

3.3. Data Collection Strategy 3.3.a. Tools and Steps to Retrieve Data

Twitter was selected as the platform because of its dominance as a tool to disseminate political information by users and because past research has shown that politically-focused memes propagate heavily on Twitter (Titlow, 2017). The tool used in this dissertation is Pulsar (https://www.pulsarplatform.com/). Pulsar is a commercial social media analytics tool which I was given access to via the Visual Social Media Lab which is directed by my supervisor, Dr Farida Vis. To further analyse the 100 most shared images and process the data, I also worked with Excel and R Studio.

Using Pulsar, only image-containing tweets that contained a predefined set of political keywords were downloaded. Data was retrieved between May 1st and June 9th following the steps. ● Insert relevant query into Pulsar (see Section 3.3.b. for exact search terms) once the ethics approval was given (See Appendix 4) ● The search was limited only to UK users, English language ● Collect the data via Pulsar organised by key phrases/hashtags ● Select the 100 most retweeted tweets ● Build a code frame on the 100 data items ● Conduct analysis of memes using the three methods outlined above (Section 3.2.)

3.3.b. Search Terms

Three key phrases and/or hashtags dictated the data collection process for this dissertation. The selected search terms corresponded to three key aspects in Theresa May's campaign which have elicited rich visual responses from Twitter users in the form of memes: 1. ‘Strong and stable’ – linked to the Conservative Party Manifesto and campaign slogan. The terms searched for were: 'strong and stable', ‘#strongandstable’, ‘strongandstable’. This search resulted in 78,518 tweets for the dates searched for. 2. ‘Weak and wobbly’ – linked to the manifesto U-turn on social care policy, otherwise known as ‘Dementia Tax’. The terms searched for were: 'weak and wobbly', ‘#weakandwobbly’, ‘weakandwobbly’. This search resulted in 22,085 tweets for the dates searched for.

3. ‘#TheresaMayGIFs’ – linked to the live ITV News appearance and interview with Jeremy Paxman. The terms searched for: ‘Theresa May GIFs’, ‘#TheresaMayGIFs’, ‘TheresaMayGIFs’. This search resulted in 16,573 tweets for the dates searched for.

3.3.c. Sample Size

In terms of sample size of tweets for each case study, the 100 most retweeted meme-containing tweets were selected. I loosely analysed the memes in order to get a sense of the suitable categories for the code frames. The 100 memes per case study were coded along the three methods: content analysis of image and text, actor type analysis, sentiment analysis of image and text. A second coder coded 20% of the total sample size, i.e. 20 data items for each code frame. The limit of 100 items per case study was imposed in order to keep the dataset at a manageable size.

3.3.d. Data Storage and Security

I had password protected access to Pulsar. When performing the search for each key phrase/hashtag, I had access to the full dataset, however most of the data remained in the tool and I only downloaded 100 tweets per case study, which I then further processed using Excel and R.

The data was stored in a University of Sheffield Google Drive system under the Gmail account of [email protected]. Further security steps were enforced by password-protecting the folders containing the data. When required, access to the data was given to the supervisor to facilitate discussions about how the research can be improved, and second coders were only given access to a small subset of the data to complete the intercoder reliability checks. The analysis of the data took place on a personal computer.

3.3.e. Ethics

Ethics approval (Application number 014268 - See Appendix 4) was obtained from the Information School Department of the University of Sheffield and this research for this project was informed by the ethical principles for social media research set out by the Association of Internet Researchers Guidelines (https://aoir.org/ethics/).

As per Research Ethics Policy Note 14, the research undertaken for this project is legal. This is in accordance with Twitter’s Terms of Service which state in part: ‘Except as permitted through the

Services, these Terms, or the terms provided on dev.twitter.com, you have to use the Twitter API if you want to reproduce, modify, create derivative works, distribute, sell, transfer, publicly display, publicly perform, transmit, or otherwise use the Content or Services.’ (Twitter, Terms of Service). Therefore, this research is legal according to Twitter as long as one of their APIs is used to collect data. Pulsar legally uses the Full-Archive Search API for Twitter which offers ‘complete and instant access to the entire archive of Twitter data’ (http://support.gnip.com/apis/search_full_archive_api/). I was given password-protected access to this tool for the purposes of this dissertation and only the data required for analysis was downloaded. I downloaded only the top 100 most retweeted tweets per case study (i.e. 300 data items in total), keeping most of the data in the tool.

This chapter covered the methods and methodology used to analyse 300 memes, namely content analysis of image and text, actor-type analysis and sentiment analysis of image and text. The results of the analysis, findings and discussion are presented in the following chapter.

Chapter 4: Findings, Analysis and Discussion

This chapter is structured into four sections. The first three sections correspond to the three case studies, namely ‘Strong and Stable’, ‘Weak and Wobbly’ and ‘TheresaMayGIFs’. The first three sections of this chapter present the findings of the analysis using the same structure: Content of memes; Posters of memes; Stance of memes; Discussion. The fourth section of this chapter compares the three case studies and discusses the intertextuality that occurs between the three case studies and the two visual motifs, namely ‘Maybot’ and ‘Laughing Theresa May’.

4.1. Findings: Strong and Stable

As explained in Section, 3.1.1., the activity of this phrase/hashtag on Twitter is derived from the Conservative slogan ‘Strong and stable leadership in the national interest’ (Official Conservatives Twitter Account, n.d.), which was repeated numerous times by the PM during the election.

This hashtag had a long and continuous activity on Twitter and it almost became synonymous with Theresa May’s campaign. Figure 1 was created in Pulsar and indicates the activity of the phrase being used on Twitter from the 1st of May, and fading after the GE. The dark blue line represents the actual number of original tweets, and the line in light blue refers to the retweets the tweets got on Twitter. The three notable peaks in the dark blue graph correspond to i) when Belgian MEP Guy Verhofstadt mocked Theresa May on Twitter for the excessive use of the phrase - on the 2nd May (Horton, May 2017), ii) the press conference in which TM was characterised as ‘weak and wobbly’ by journalist Michael Crick - on the 22nd May (see Section 3.1.2.), and iii) the election itself on the 9th of June.

Figure 1. Pulsar activity for SAS

4.1.a. Content of Memes

For this case study, the content of the tweet, and in turn the meme shared in the tweet, consisted of different types of media. Given the focus of this dissertation, it was important to distinguish which memes referred directly to Theresa May, which of them were addressed at Jeremy Corbyn, which of them included both, and which memes simply included the phrase or hashtag ‘Strong and Stable’ without mentioning a party leader. Figure 2 illustrates this distribution: out of 100 memes, 62% of them included a reference only to Theresa May, 4% referred only to Corbyn, 5% of them addressed both party leaders, and 39% of the memes included the phrase without mentioning the leader. The next parts of the analysis excluded the memes that only addressed Jeremy Corbyn.

The next stage of the analysis was concerned with the themes/issues that the memes, along with the text accompanying the meme, referred to. For this a code frame was developed containing seven categories. Figure 3 shows the breakdown of the issues Twitter users addressed when sharing memes containing the phrase or hashtag ‘Strong and Stable’. The most common theme referred to the leadership of the PM and the Conservative Party in general. The following themes that emerged in the tweets were, in descending order: the Conservative Party manifesto and policies (20.8%), the election campaign (19.8%), a call to vote/support Theresa May (11.5%), personal attacks at her (either referring to her appearance or personality) (9.4%), and, lastly, the competition between her and Corbyn (4.2%). The tone permeating these tweets was in most cases, except for the ‘Call to Vote’ category, of a critical nature, pointing out to the failures of the Conservative Party or Theresa May to address certain issues, ridiculing tones as well as sarcastic commentary towards her and her party.

4 The tweets in this case study contained different types of memes in different formats. Figure 4 indicates the different types of memes, with a dominating type being the ‘Images with text’ category (26%), followed by ‘Images of newspapers, articles or parts of text’ (21.9%), followed by ‘Images’ -

4 Note: the percentages in Figure 3 add up to 100.1% because the percentages were rounded up to have only one decimal point

that did not depict Theresa May (18.8%), ‘Video’ (16.7%), ‘Images’ - that had her the centre of the photo (10.4%), and, lastly, GIFs (6.3%) that did not depict her and represented amusing scenes from movies, series or cartoons.

4.1.b. Posters of Memes

In terms of the posters of the ‘Strong and Stable’ memes, Figure 5 indicates the types of users that shared content using this phrase/hashtag. Most of them were political actors (26%) that criticised the PM or an aspect of her campaign, as well as official content tweeted by the Conservative Party calling voters to support May in the GE; among these political actors there was also content shared by the Labour Party attempting to undermine May’s credibility as a leader and using the phrase ‘strong and stable’ in a sarcastic tone. The next largest category included Members of the public (19%) who voiced their concern towards certain aspects of May's campaign, or expressed their support towards her. The next categories of users included, in descending order, Organisations (15.6%), Journalists (13.5%), Celebrities (8.3%), Bloggers (4.2.), Corbyn-Fan Accounts (3.1%), Researchers, Media, Activists (each 2.1.%) and Activists, Mock Theresa May Accounts and Other (each 1%).

In terms of the political leaning of the users that shared this content, Figure 6 illustrates that the majority of users (44.8%) were left-leaning, and only 15.6% were right-wing users. For the rest of 39% it was impossible to determine their political leaning.

The analysis investigated next how many tweets came from official party sources. Figure 7 demonstrates that 13.5% of the 15.6% right-wing users were Conservative Party official accounts (i.e. the Conservative Party itself @Conservatives, or Theresa May @theresa_may). Therefore, the percentage of Twitter users that identified as right-wing and were not party official accounts is 2.1%. Meanwhile, 41.7% of the users were identified as left-leaning and for 39.6% of users it was impossible to determine the political leaning.

4.1.c. Evaluative stance of memes

Three categories were developed to categorise the stance of the meme and content: ‘Critical’, ‘Supportive’, ‘Neutral’. The categories were applied in two stages. First, the memes were coded in isolation of the text. Figure 8 indicates that 33.3% of the SAS memes were critical, 13.5% of the memes were supportive of Theresa May and most of the memes (53.1%) were neutral (neither critical nor supportive).

In the second stage of the analysis the same code frame was applied to both the meme and the text. The reason for this was to see whether the meaning and intention behind the memes changed when viewed with the text. We can see that the proportion of content that has a ‘Neutral’ stance is drastically reduced from the 53.1% (Figure 8) to only 9.4% (Figure 9). Moreover, the proportion of content with a ‘Critical’ stance increases from 33.3% (Figure 8) to 76% (Figure 9).

4.1.d. Strong and Stable: Discussion

The following insights emerge from the analysis of this case study. First, this case study fits with Lynch & Dolan (2013), Campus’s (2013) and Harmer et al. (2016) research that a key component of successful UK politicians is the ability to displaying ‘strong’ qualities. The decision to choose this particular slogan in the first place is a direct confirmation of this finding, as Theresa May sought to position herself as the ‘safest choice’ for the British public. However, as Parry (2015) highlighted, the excessive repetition of the SAS slogan led to the portrayal of Theresa May as a one-dimensional leader, that couldn’t manage to capture the public’s imagination in terms of creating supportive visual content. The evidence found in this study supports this view, as the tweets and memes themselves portrayed Theresa May as a robot - also in alignment with the Maybot visual motif.

Second, the memes contained all three categories of evaluative stances, namely critical, supportive and neutral. This was the only case study in which official sources from the Conservative Party appeared as a type of user, performing what Jensen (2017) called a ‘call to action’. However, as detailed in Section 2.1., the party is using Twitter more as a tool to broadcast their message, rather than to engage with users (Jensen, 2017).

Third, while the Conservative Party did integrate social media in their campaign and shared memes, specifically images with superimposed text, their message and content delivery was repetitive. The analysis shows that the same image was repeated over and over again. Twitter users were sensitive to that repetition and, as a result, appropriated that expression and used it for criticism and not appraisal. The Conservative Party lost control of the narrative and their slogan and core message was used against them through the use of memes and visual representations.

4.2. Findings: Weak and Wobbly

As discussed in section 3.1.2, this case study is linked to Michael Crick’s characterisation of TM and her proposed social care policy as ‘weak and wobbly’. As can be seen in the Pulsar visualisation (Figure 10), this hashtag also had a continuous activity on Twitter, as the phrase was picked up by the press, members of the public and the Labour party. Figure 10. Pulsar activity for WAW

4.2.a. Content of Memes

For this case study, it was important to distinguish which memes referred directly to or included Theresa May alongside the phrase/hashtag ‘weak and wobbly’. Figure 11 illustrates that out of 100 memes, 72% of them included a reference only to Theresa May, 22% of the memes included the phrase without a mention of a party leader, 4% of them addressed both party leaders, and 2% only referred to Corbyn. The next parts of the analysis excluded the memes that only addressed Jeremy Corbyn.

The next step of the analysis revealed the kinds of themes that were contained in the memes. Figure 12 reveals that almost half of the data (46.9%) referred to May's policies and manifesto. The next issues presented in the tweets related to Theresa May's refusal to participate in televised debates and confronting Jeremy Corbyn directly (15.3%), the Conservative Party campaign overall (11.2.%), personal attacks at the PM that were using unflattering pictures to undermine her authority (7.2%), and lastly, stabs at her leadership abilities which were characterised as ‘weak and wobbly’ and the opposite of ‘strong and stable’ (4.1%).

The fact that most of the tweets referred to policies and the manifesto was expected, given that the phrase appeared as a response to a social care policy change, only four days after the launch of the Conservative Party Manifesto. However, it was important to dig further into that 46.9% category, and be more specific about the exact policies the tweets referred to (i.e. counting the number of times a policy was mentioned). Figure 13 confirms that most of the tweets used the phrase ‘weak and wobbly’ in relation to the Conservative social care policy (n=23), either by criticising Theresa May for her ‘U-turn’ or highlighting their frustration with the policy. Other policy issues mentioned in the tweets included, in descending order: the manifesto in general (n=10), Brexit (n=6), the NHS (n=3), police cuts, school lunches (n=2), income tax (n=1) and immigration (n=1).

The memes that were shared using the ‘weak and wobbly’ phrase consisted of different types and formats (Figure 14). As was observed in the SAS case study, most memes were ‘Images with text’ (38.8%), followed by ‘Video’ (25.5%), ‘Images of text - newspapers, articles, screenshots of text, leaflets’ (21.4%), ‘Images of Theresa May without text’ (7.1%) and other ‘Images’ (4.1%).

4.2.b. Posters of Memes

In terms of the Twitter users that shared content using the ‘Weak and wobbly’ phrase/hashtag (Figure 15), most of them represented Members of the public (24.5%), followed by Journalists (16.3%), Organisations (12.2%), Jeremy Corbyn Fan accounts (10.2%), Political actors (9.2%), Bloggers (7.1%), Media (3.1%), Activists (2%) and Accounts mocking Theresa May (1%).

In terms of the political leaning of the users, given the inherently critical tone of the phrase, Figure 16 shows that most users were left-leaning (74.1%), 27.6% were neutral (i.e. they did not explicitly mention a political leaning in their bio/photo/username) and only 1 user was coded as right- leaning. Unlike the ‘strong and stable’ case study, none of the content came from official Conservative Party sources.

4.2.c. Evaluative stance of the meme

To determine the evaluative stance of the ‘weak and wobbly’ memes, the same three-category code frame was applied. First, the memes were analysed in isolation of the text. Figure 17 indicates that 53.1% of the memes were critical of Theresa May and/or other aspects of her campaign. Also, 46.9% of the memes, on their own, appeared to be neutral. None of the memes had a ‘Supportive’ stance.

When the same code frame was applied to the meme as well as the text of the tweet, the proportions in terms of evaluative stance changed. In Figure 18 the percentage of content with a ‘Critical’ stance is 81.6% (a 28.5% increase from Figure 17). At the same time, the proportion of content with a ‘Neutral’ stance dropped to 18.4% (a 28.5% decrease from Figure 17).

4.2.d. Weak and Wobbly: Discussion

The following insights emerge from the analysis of this case study. First, this case study and the behaviour of the hashtag/phrase online illustrates a situation in which something that took place in traditional media was appropriated on social media, taking a life of its own and amplifying it. What could have been an unplanned off-handed remark, became more sharp and intense. This raises interesting questions about how social media can appropriate and redirect critical reflections on a political candidate. In many ways, this case study represents supports what Ross & Rivers (2017) describe as the use of deligitimisation strategies to undermine Theresa May and her credibility (Section 2.3.b).

Second, the ‘weak and wobbly’ phrase was associated by Twitter users with austerity and ‘small government’ policies, inferring that genuine ‘strong and stable’ leadership would require the opposite of that, namely, that the government take a direct role in assuming social responsibilities towards the elderly, healthcare, policing or school policies. Moreover, although the phrase was originally used only in reference to social care policy, it quickly became apparent that other Conservative policies and campaign elements were available and suitable for criticism using the same phrase.

Third, this case study highlights the one-sidedness of the conversation, in that no official Conservatives or any apparent (except one) Conservatives sought to defend their candidate and policies. There are two ways to read this: one, it was a deliberative strategy not to engage in order to avoid providing more ammunition and enabling the phrase to go even more viral, or two, that they were unable or unwilling to defend their position against the ‘weak and wobbly’ claims. Either way, this allowed a particular narrative to dominate and for a particular slogan to be amplified in relation to Theresa May.

4.3. Findings: TheresaMayGIFs

As explained in section 3.1.3. the activity of the hashtag can be explained in relation to the TV show ‘Battle for Number 10’ and the Jeremy Paxman interview on the 29th of May. In contrast with the two other case studies, this hashtag had a limited timespan, as can be seen in the Pulsar visualisation (Figure 19).

Figure 19. Pulsar activity for TMG

4.3.a. Content of Memes

Given the hashtag of this case study, the assumption was that, although the memes contained in the tweets may consist of several types of media, most of the memes would be GIFs. Figure 20 illustrates that indeed 90% of the memes were GIFs, 7% were images (with and without text), 2% were videos and 1% was text. The next parts of the analysis only focused on memes in GIF format.

As in the previous case studies, it was imperative to narrow down the focus of the memes and exclude memes that referred only to Jeremy Corbyn. Figure 21 shows that 38.9% of the GIFs referred only to Theresa May, 53.3% did not include the name of either party leader and only used the hashtag on its own, 5.6% referred to both leaders and 2.2% referred only to Corbyn. The next parts of the analysis excluded the GIFs that only addressed Jeremy Corbyn.

Given the dynamic nature of GIFs as animated clips, able to capture more complex emotions than static images, a code frame was developed to reveal the actions depicted in the GIFs. These actions naturally emerged in the initial stages of observing the data. The 44 Twitter GIF categories available to Twitter users were also used as inspiration. Figure 22 illustrates the process of inserting a GIF into a tweet and 6 of the 44 available categories.

Figure 22. Twitter Snapshot of how to select a GIF for a tweet

Figure 23 shows that most of the GIFs (28.4%) depicted a fall. Although quite an extensive range of emotions was extracted from the GIFs - with 9 categories corresponding to different feelings and situation - the ‘Other’ category in this instance holds second position in terms of results (18.2%). Despite this, the coding still allowed for a granular categorisation of the memes into eight other categories, arranged here in descending order: Disbelief (12.5%), Fire/Destruction (11.4%), Hiding, (10.2%), Backfire (5.7%), Regret (4.5%), Fight (2.25%) and I Don’t Know (2.25%).

Content analysis was conducted on the text and the GIFs to reveal the topics they covered (Figure 24). As expected, given that the hashtag appeared as a response to the Live Q&A event and the Jeremy Paxman interview, most of the tweets were a commentary on the PM’s performance that night (51.1%) - referring to her interaction with voters, her inability to answer their questions directly as well as the supposed reaction of the Conservative party and supporters (generally of regret and desire to disassociate). The next topics addressed her manifesto and policies (13.6%), the election campaign in general (11.4%), and the competition between May and Corbyn (3.4%). There was also a significant portion of tweets whose text simply included the hashtag ‘#TheresaMayGIFs’ and a GIF, without giving any further explanation as to what aspect in particular the content referred to (13.6%).

4.3.b. Posters of Memes

Figure 25 indicates the types of Twitter users that shared GIFs using the #TheresaMayGIFs hashtag. Most of them were Members of the public (65.9%) who used the GIFs as a way to comment on and engage with Theresa May’s performance, followed by Journalists (9.1.%), Celebrities (5.7%), Bloggers (3.4%), Researchers (3.4%) and Twitter accounts set up specifically to mock Theresa May (2.3%). The smallest categories of 1.1% each were Political actors, Activists and Fan accounts for Jeremy Corbyn.

In terms of the political leaning of the accounts that engaged with the #TheresaMayGIFs hashtag, Figure 26 illustrates that for most of the accounts it was not clear whether they were left or right- leaning (71.6%), 28.4% were left-leaning and none of the accounts self-identified as right-leaning. No official party accounts engaged in sharing content.

4.3.c. Stance of Memes

In terms of the evaluative stance of the GIFs, the same three-category code frame was applied in isolation to the GIF, and then to the GIF and text as a whole. The first round of coding (Figure 27) revealed that 55.7% of the GIFs were critical either of Theresa May directly, or of any of the issues reflected above in Figure 24 in Section 4.3.a.

When the code frame was applied to the text and the GIF, the results were similar to the other two case studies, in that the percentage of content with a ‘Neutral’ stance reduced drastically, from 44.3% (Figure 27) to 5.7% (Figure 28). Moreover, the percentage of content with a ‘Critical’ stance increased to 94.3% (Figure 28) from 38.6% (Figure 27).

4.3.d. TheresaMayGIFs: Discussion Three insights emerge from analysing this case study. First, the use of GIFs in relation to the live Q&A and interview supports Giglietto’s (2014) findings about the ‘second-screen experience’ according to which people often use social media while they watch TV. This form of engagement was evident in this case study, as users chose to experience the TV moment in a different way online, via the use of GIFs and the #TheresaMayGIFs hashtag to express their concerns, feelings and attitudes. This interactive experience allowed users to engage with other users on the topic and participate in the active remixing of digital objects and meme-sharing.

Second, the findings suggest that people’s desire to engage in political debates, they want to voice their frustrations and have those recognised and shared by others. It was evident that people sought a way to vent their political frustrations and, in this case, social media provided that opportunity for venting and biofeedback. As elaborated in Section 2.1., these findings supports Lilleker and Jackson’s (2010) argument about social media creating a participatory architecture that facilitates interactions between campaigns and supporters that otherwise do not exist.

Third, this case study strengthens the literature that describes the practice of sharing reaction-GIFs (Section 2.3.b.) in response to political events and actions, as they capture emotional and bodily reactions and are meant to be inserted into the flow of conversation (Huber, 2015). This reactive practice made the conversation dynamic and interactive and allowed for the creation of an online space where people could express their views in a specific visual format.

In conclusion, the activity of the three case studies overlaps (Figure 29), with SAS displaying the most longevity (dark blue), followed by WAW (grey) and lastly TMG (pink). The period between May 22nd until June 9th manifested intense Twitter activity under these three case studies and supports the argument that social media plays an important part in political campaigns and elections (Miller, 2018; Jensen, 2017).

Figure 29. Pulsar activity for all case studies, SAS, WAW and TMG

4.4. Comparing Case Studies

As explained in the Introduction and in Sections 3.4. and 3.5., two additional visual motifs were identified in this dataset to help understand the content and stance of the memes. During the coding and analysis of the memes it became clear that there is an element of intertextuality associated with the practice of meme-sharing, as certain references can spill over and mix with other key phrases/hashtags creating multi-layered visual interpretations. This indicates that these memes are being shared by people ‘in the know’, and that these motifs and case studies are a shorthand for certain characterisations, expressions, feelings (Shifman, 2013; Milner, 2013). There is an inherent ‘layered-ness’ taking place, in which the visual representation of Theresa May is multidimensional in terms of the references chosen to comment on her as a politician and campaign. To illustrate this point, Table 2 shows that all three case studies include a reference to another visual motif, with

WAW containing one tweet that has two references (Figure 30-31). Some examples of memes that highlight intertextuality are also given in Images 6a-g.

Table 2: Comparison between the three case studies

Case Reference to other case studies/visual motifs study

SAS

Image 6a - Shared on Twitter (22 May 2017); Image 6b - Shared on twitter (16 May 2017)

Figure 30: References to other case studies/visual motifs in the SAS case study

WAW

Image 6c - Shared on Twitter (22 May 2017); Image 6d - Shared on Twitter (30 May 2017)

Figure 31: References to other case studies/visual motifs in the WAW case study

TMG

Image 6e - Shared on Twitter (29 May 2017); Image 6f - Shared on Twitter (29 May 2017)

Image 6g - Shared on Twitter (30 May 2017)

Figure 32: References to other case studies/visual motifs in the TMG case study

This chapter presented the findings for each of the case studies, namely ‘Strong and Stable’, ‘Weak and Wobbly’, and ‘Theresa May GIFS’. The findings of the analysis were presented using the same structure: Content of memes; Posters of memes; Stance of memes; Discussion. The fourth section of this chapter discussed the intertextuality that occurred between the three case studies and the two visual motifs, namely ‘Maybot’ and ‘Laughing Theresa May’, highlighted the multi-layered visual interpretation that takes place within the practice of meme sharing. The following and final chapter will present the concluding remarks, discuss the limitations of the research and provide direction for potential future research.

Chapter 5: Conclusion

This chapter concludes the findings from its research in previous chapters and is structured into three parts. First, it revisits the original research questions and research objectives that the dissertation set out to answer and meet, and evaluates the extent to which those were satisfactorily addressed. Second, it discusses the limitations of the research. Finally, it elaborates on potential directions for the research of political memes moving forward.

5.1. Review of Research Questions and Research Objectives - Summary of main conclusions and findings

The aim of this dissertation was to better understand the ways in which Twitter users utilised memes during the 2017 GE to respond and engage with Theresa May’s campaign and performance as a politician and specifically, as a female politician.

To do so, three case studies were selected, each corresponding to a key aspect of her campaign. These were ‘Strong and Stable’ – linked to the Conservative Party Manifesto and campaign slogan; ‘Weak and Wobbly’ – linked to the manifesto U-turn on social care policy; and, ‘TheresaMayGIFs’ – linked to the live ITV News appearance on the ‘Battle for Number 10’ show’ and the interview with Jeremy Paxman.

For each case study, the 100 most retweeted tweets containing images were selected for analysis using three methods: content analysis of image and text, actor type analysis, and sentiment analysis of image and text. Through the analysis of these memes, this dissertation aimed to answer three research questions and to meet three research objectives.

RQ1 & RO1: What is the content of the memes used on Twitter in the three case studies? - Understand the kinds of content depicted in these memes and the subject or object they refer to. This question was answered and the objective was met. Through content analysis, the subject depicted in the memes was either solely Theresa May or Jeremy Corbyn, both of them at the same time or none of them with the phrase being used on its own. Moreover, through CA, the issues that were referred to in the memes and the text were identified.

RQ2 & RO2: What types of users share those memes? - Understand the type of Twitter accounts/users engaged in the sharing of political memes of Theresa May during the GE. This referred to the category of user (i.e. media, organisation, member of the public, etc.) as well as the type of political leaning of the account (i.e. left/right-leaning, neutral, official Labour, official Conservative sources). This question was answered and the objective was met. Through actor type analysis, the categories of Twitter users were identified and coded accordingly. In addition to this, the political leaning of the accounts was also determined based on their Twitter profile information.

RQ3 & RO3: What is the evaluative stance of the memes in relation to May’s political leadership? - Understand how people positioned themselves with reference to Theresa May (critical / supportive / neutral). This question was answered and the objective was met. Through SA on image and text, the stance of the memes was determined in terms of it being critical, supportive of neutral of Theresa May or an aspect of her policies, campaign or leadership.

Three main findings emerged from the analysis. First, in two of the three case studies (SAS and WAW), Theresa May’s policies and manifesto were revealed to be the most referred to issues by Twitter users, while Theresa May’s TV performance was the most frequent issue in the TMG case study. Second, in two of the three case studies (WAW & TMG), members of the public were the most active category of users, with the third one being political actors (SAS). Lastly, all case studies revealed that, when analysed both in isolation of the text as well as alongside the text, the memes were critical of Theresa May or an aspect of her policies, campaign or leadership.

On reflection of these findings, four broader implications seem to have been generated from this research.

First, memes can have a political orientation, and this has the potential to be negative, while messages and slogans can be re-appropriated through memes. Even though a candidate may try to push a positive message for their campaign, this exact message can be used as a form of critique against them. The viral nature of Twitter allows that negative message or critique to gain momentum. In addition, innocent acts or mistakes can also be amplified through the use of memes,

by their viral nature they can take on a life of their own, and in the process, undermine the credibility of the leader.

Second, these findings suggest that memes are a powerful form of political expression, that they are often used by people to vent points and make commentary (albeit mostly negative comment and critique). This in turn implies that it is very difficult for parties to control the narrative online and that traditional methods of staying ‘on message’ may backfire. This is because messages can be appropriated to show the opposite message than intended (as in this case, Theresa May was depicted as ‘weak and wobbly’, and not ‘strong and stable’).

Third, political parties could consider more clever ways and be savvier when it comes to ways in which they deal with Twitter and memes. If there is an efficient way to utilise memes in a political campaign, then it is most likely as an attack on opponents, instead of as a means to support one’s own candidate. Moreover, it is more likely that the use of memes is effective if it does not come from the Party itself, but from supporters of the Party in order for it to have that grassroots legitimacy.

Lastly, traditional forms of political campaigning and communication, which may imply repeating the same line or slogan over and over again, are not effective when it comes to certain social media avenues, and particularly Twitter and memes. In many ways, the relationship between social media and politics is a double-edged sword. On the one hand, politicians can become more self-conscious, controlled and sterilised in their public appearances out of fear of being ridiculed, or their words or actions taken out of context. On the other hand, social media users can hold their leaders accountable, and actions such as repeating the same slogan excessive amount of times in a speech will not go unnoticed.

5.2. Limitations of the research

Although this research gives rise to a series of interesting insights on the use of memes in the 2017 GE, there are at least three limitations involved with this research.

First, CA is a purely descriptive method, that describes what is there, but may not reveal the underlying motives for the observed sharing practices. Similarly, for SA, the interpretation of sentiment is subjective, and issues such as sarcasm are challenging to determine. Both methods are

subjective and developing code frames and manually coding is very time-consuming, which means that only relatively small samples can be included.

Second, the observed trends in meme-sharing on Twitter are not an accurate reflection of reality or of the UK voting population as a whole. This is a well-documented criticism of research that deals with Twitter data, however it is not the intention of this research to extrapolate the findings to the general population, but merely depict practices that are happening online and the potential insights this has on the use of memes in elections.

Lastly, the metric that determined the most popular tweets for each of the case studies was based on the number of retweets, which as Driscoll and Walker (2014) argue may not necessarily be the most accurate metric to illustrate reach and influence online.

5.3. Suggestions for Future Research

Despite these limitations, this research suggests several future research projects that can provide further insights into how memes are used a form of political expression during election campaigns. One, three individuals research projects could focus on each of the three case studies, and using discourse analysis as a method, configure a 360-degree view of those phenomena. Two, a potential quantitative study that tries to map the effect of memes on voter preference or undecided voters, through surveys or other methods. Third, a sociological study on why people are motivated to use memes as a form of expression in politics. Lastly, it would be interesting to understand memes as a wider deliberative mechanism within the public sphere.

- The End -

References

Andrews, P. (5 July 2017) EVERY DAY CAN BE ED BALLS DAY IN UK POLITICS FANDOM, Discover Society, http://discoversociety.org/2017/07/05/every-day-can-be-ed-balls-day-in-uk-politics- fandom/

Anstead, N., O’Loughlin, B. (2010) The Emerging Viewertariat: Explaining Twitter Responses To Nick Griffin’s Appearance On BBC Question Time, PSI Working Paper, University of East Anglia, pp. 1-21, Available at https://static1.squarespace.com/static/566d81c8d82d5ed309b2e935/t/567ab24b1c12101fd291 a3c7/1450881611076/Anstead_OLoughlin_BBCQT_Twitter_Final.pdf

BBC (18 April 2017) Theresa May's general election statement in full, BBC, Available at http://www.bbc.co.uk/news/uk-politics-39630009

BBC (22 May 2017) The Guardian view on the social care debacle: weak and wobbly, The Guardian, https://www.theguardian.com/commentisfree/2017/may/22/the-guardian-view-on-the-social- care-debacle-weak-and-wobbly

Belam, M. (29 May 2017) #TheresaMayGIFs: hashtag mocks PM's performance on TV, The Guardian, Available at https://www.theguardian.com/politics/2017/may/29/theresa-mays-tv- performance-mocked-with-theresamaygifs-hashtag

Benkler, Y. (2006) The Wealth of Networks: How Social Production Transforms Markets and Freedom. New Haven: Yale University Press.

Blackmore, S. (1999). The meme machine. Oxford, England: Oxford University Press.

Broersma, M. (2015) ‘Hot Dog Politics: Why comfort food makes politicians uncomfortable’, p.23 in Jackson, D, Thorsen, E. (2015), ‘UK Election Analysis 2015: Media, Voters and the Campaign, The Centre for the Study of Journalism, Culture and Community

Bryman, A. (2012) Social Research Methods. (4th Ed.). Oxford, England: Oxford University Press

Burroughs, B. (2013), FCJ-165 Obama Trolling: Memes, Salutes and an Agonistic Politics in the 2012 Presidential Election, The Fibreculture Journal, pp. 258-277.

Bush, S. (3 May 2017) There is one place where Labour's campaign is strong: radio, New Statesman, Available at http://www.newstatesman.com/politics/june2017/2017/05/there-one-place-where- labours-campaign-strong-radio

Campus, D. (2013) Women Political Leaders and the Media. New York: Palgrave Macmillan. Charles, A. (2015) ‘The politics of social media’, p. 67 in Jackson, D, Thorsen, E. (2015), ‘UK Election Analysis 2015: Media, Voters and the Campaign, The Centre for the Study of Journalism, Culture and Community

Conservative Manifesto (2017) https://www.conservatives.com/manifesto / https://issuu.com/conservativeparty/docs/ge2017_manifesto_a5_digital/1?ff=true&e=16696947/ 48955343

Conservatives Twitter Account, https://twitter.com/Conservatives

Crace, J. (6 November 2016) Theresa struggles to take back control – from her own Maybot, The Guardian, Available at https://www.theguardian.com/politics/2016/nov/08/theresa-may- struggles-take-back-control-maybot-india-brexit

Dawkins, R. (1976). The selfish gene. Oxford, England: Oxford University Press.

Dathan, M. (3 April 2015) Who won the leaders' TV debate?, The Independent, Available at http://www.independent.co.uk/news/uk/politics/generalelection/who-won-the-leaders-tv- debate-nicola-sturgeon-according-to-these-labour-people-10153695.html

Driscoll, K., Walker, S. (2014) Big Data, Big Questions| Working Within a Black Box: Transparency in the Collection and Production of Big Twitter Data, International Journal and Communication, 8, pp. 1745-1764

Elgot, J., Martinson, J. (18 April 2017) Theresa May rules out participating in TV debates before election, The Guardian, Available at https://www.theguardian.com/politics/2017/apr/18/theresa- may-rules-out-participating-in-tv-debates-before-election

England, C. (18 May 2017) Conservative manifesto 2017: All you need to know about the Tories’ election pledges, The Independent, http://www.independent.co.uk/news/uk/politics/conservative-manifesto-2017-all-need-know- key-points-tory-election-policies-theresa-may-a7743001.html

Eppink, J. (2014) A brief history of the GIF, Journal of Visual Culture, 13:2, pp. 298-306, DOI 10.1177/1470412914553365

Freelon (2010), ReCal2, http://dfreelon.org/utils/recalfront/recal2/ Gibson, R. K. (2015) Party Change, Social Media and the Rise of ‘Citizen-Initiated’ Campaigning.” Party Politics 21 (2): 183–97. doi:10.1177/1354068812472575.

Giglietto, F., Selva, D. (2014) Second Screen and Participation, Journal of Communication, 64, pp. 260-277, doi:10.1111/jcom.12085.

Gil de Zúñiga, H., Jong, N., Valenzuela, S., (2012) ‘Social media use for news and individuals' social capital, civic engagement and political participation’, Journal of Computational Media, Communication 17:3, pp. 319–336. http://dx.doi.org/10.1111/j.1083-6101.2012.01574.x.

Goodman, P. (20 April 2017) Crosby was in charge of Cameron’s campaign. Will he be the master of May’s?, Conservative Home, http://www.conservativehome.com/thetorydiary/2017/04/crosby- was-in-charge-of-camerons-campaign-will-he-be-the-master-of-mays.html

Gorgovenko, K., Taylor, N. (2016) Politics at Home: Second Screen Behaviours and Motivations During TV Debates, Proceedings of the 9th Nordic Conference on Human-Computer Interaction, 22, pp. 1-10.

Graham, C. (30 May 2017) #Theresamaygifs: Prime Minister mocked on social media for debate performance, The Telegraph, Available at http://www.telegraph.co.uk/news/2017/05/30/theresamaygifs-prime-minister-mocked-social- media-debate-performance/

Guardian Live Feed (30 May 2017) Labour and Tory leaders interviewed by Jeremy Paxman - as it happened, The Guardian, Available at https://www.theguardian.com/politics/blog/live/2017/may/29/paxman-interview-corbyn-may- sky-general-election-paxman-interviews-may-and-corbyn-politics-live

Hall, S. (1973) Encoding and Decoding in the Television Discourse, Birmingham: Centre for Contemporary Cultural Studies.

Harmer, E., Savigny, H., Ward, O. (2016) ‘Are you tough enough?’ Performing gender in the UK leadership debates 2015, Media, Culture and Society, pp. 1-16, DOI: 10.1177/0163443716682074

Harris, S.A. (30 May 2017) Theresa May Gifs Are Maybe The Best Thing To Come From Her Jeremy Paxman Interview, The Huffington Post, Available at http://www.huffingtonpost.co.uk/entry/theresa-may-gifs-paxman- interview_uk_592d1d50e4b0df57cbfcfee9

Hayes, A. F., & Krippendorff, K. (2007). Answering the call for a standard reliability measure for coding data, Communication Methods and Measures (1), pp. 77-89.

Hindman, Matthew S. (2009) The Myth of Digital Democracy. Princeton: Princeton University Press.

Hines, N. (08 June 2017) How Theresa May’s ‘Strong and Stable’ Pitch Went Weak and Wobbly, Daily Beast, Available at http://www.thedailybeast.com/how-theresa-mays-strong-and-stable- pitch-went-weak-and-wobbly

Horton, H. (02 May 2017), EU Brexit representative trolls Theresa May on Twitter over her use of 'strong and stable, The Telegraph, Available at http://www.telegraph.co.uk/news/2017/05/02/eu-brexit-representative-trolls-theresa-may- twitter-use-strong/

Huber, L. (2015) ‘Remix culture and reaction GIF’, GNOVIS, Available at http://www.gnovisjournal.org/2015/02/25/remix-culture-the-reaction-gif/

Hughes, M.A., Garrett D. E. (1990) Intercoder Reliability Estimation Approaches in Marketing: A Generalizability Theory Framework for Quantitative Data. Journal of Marketing Research. Vol. 27, No. 2 (May, 1990), pp. 185-195

Isasun (2 October 2013) The German Election on Twitter, Twitter Blog, Available at https://blog.twitter.com/official/en_us/a/2013/the-german-election-on-twitter.html

Jiang et al (2017) http://cmci.colorado.edu/idlab/assets/bibliography/pdf/Jiang2017.pdf

Kelemen, M. (21 February 2012) Twitter Diplomacy: State Department 2.0, NPR, Available at http://www.npr.org/sections/alltechconsidered/2012/02/21/147207004/twitter-diplomacy- state-department-2-0

Khan-Ibarra, S. (13 November 2014) The Case For Social Media and Hashtag Activism, HuffPost, Avialble at http://www.huffingtonpost.com/entry/the-case-for-social-media_b_6149974.html

Knobel, M., & Lankshear, C. (2007). Online memes, affinities, and cultural production. In M. Knobel & C. Lankshear (Eds.), A new literacies sampler (pp. 199-227). New York, N.Y.:Peter Lang

Lilleker, D., Nigel, J. (2013) Political Campaigning, Elections and the Internet:Comparing the US, UK, France and Germany. New York: Routledge.

Lombard, M. (2010) Practical Resources for Assessing and Reporting Intercoder Reliability in Content Analysis Research Projects, MatthewLombard.com, Accessed 15 May 2017 from http://matthewlombard.com/reliability/

Lotan, H., Graeff, E., Ananny, M., Gaffney, D., Pearce, I., Boyd, D. (2011) The Revolutions Were Tweeted: Information Flows During the 2011 Tunisian and Egyptian Revolutions, International Journal of Communication 5, Feature, pp.1375–1405

Lynch, T., Dolan, K. (2013) It takes a survey: understanding gender stereotypes, abstract attitudes, and voting for women candidates. American Politics Research. Epub ahead of print 13 September. DOI: 10.1177/1532673X13503034.

Merrick, R. (31 May 2017) Theresa May denies refusing to debate Jeremy Corbyn because she is 'frightened' of losing on live TV, The Independent, http://www.independent.co.uk/news/uk/politics/theresa-may-election-2017-debate-jeremy- corbyn-frightened-bbc-live-tv-leaders-party-a7765316.html

Miller, C.C. (7 November 2008) How Obama’s Internet Campaign Changed Politics, The New York Times Blog, Available at https://bits.blogs.nytimes.com/2008/11/07/how-obamas-internet- campaign-changed-politics/

Milner, R. (2013) Pop Polyvocality: Internet Memes, Public Participation, and the Occupy Wall Street Movement, International Journal of Communication, 7, pp. 2357-2390.

Neuendorf, K. (2002) The Content Analysis Guidebook, Sage Publications: London.

Oakley, N. (9 June 2017) Best Theresa May memes, jokes and funniest gifs as snap general election backfires spectacularly for Prime Minister, The Daily Mirror, Available at http://www.mirror.co.uk/news/politics/funniest-theresa-memes-gifs-snap-10589663

Odell, M., Mance, H. (2017, May 18) May launches Conservative Manifesto – as it happened, Financial Times, Available at http://blogs.ft.com/westminster/liveblogs/2017-05-18/

Pang, B., Lee, L. (2008) Opinion mining and sentiment analysis, Foundations and Trends in Information Retrieval, 2(2-1), pp. 1-135

Parry, K. (2015) ‘#RegisterToVote: picturing democratic rights and responsibilities on Twitter’ p. 88, in Jackson, D, Thorsen, E. (2015), ‘UK Election Analysis 2015: Media, Voters and the Campaign, The Centre for the Study of Journalism, Culture and Community

Pickard, J. (6 November 2016) When politics and social media collide, Financial Times, Available at https://www.ft.com/content/27a7d6c8-702f-11e6-a0c9-1365ce54b926

Procter, R., Vis, F., Voss, A. (2013) Reading the riots on Twitter: methodological innovation for the analysis of big data, International Journal of Social Research Methodology, 16:3, pp. 197-214, DOI: 10.1080/13645579.2013.774172

Pulsar, https://www.pulsarplatform.com/

Ratcliffe, R. (22 April 2015) Milifandom soars with Twitter backing for Labour leader Ed Miliband, The Guardian, Available at https://www.theguardian.com/politics/2015/apr/22/milifandom- soars-with-twitter-backing-for-labour-leader-ed-miliband

Reddy, S. (2016) Introducing GIF search on Twitter, Twitter Blog, https://blog.twitter.com/official/en_us/a/2016/introducing-gif-search-on-twitter.html

Rintel, S. (2013) Crisis memes: the importance of templatability to internet culture and freedom of expression, Australasian Journal of Popular Culture, 2:2, pp. 253, 271, doi: 10.1386/ajpc.2.2.253_1

Ross, A. S., Rivers, D. J. (2017) Digital cultures of political participation: Internet memes and the discursive delegitimization of the 2016 U.S Presidential candidates, Discourse, Context and Media 16, pp. 1-11.

RT UK (22 May 2017) May hits back at "weak and wobbly" criticism, YouTube, https://www.youtube.com/watch?v=bU174tye5qw

Scott, W. A. (1955) Reliability of Content Analysis: The Case of Nominal Scale Coding, The Public Opinion Quarterly, 19:3, pp. 321-325.

Shifman, L. (2013) Memes in a Digital World: Reconciling With a Conceptual Troublemaker, Journal of Computer-Mediated Communication. 18, pp. 362–377.

Simons, N. (22 May 2017) Theresa May Performs Unprecedented ‘Weak And Wobbly’ Manifesto U- Turn On ‘Dementia Tax’ Policy, Huffington Post UK, http://www.huffingtonpost.co.uk/entry/theresa-may-performs-screeching-u-turn-on-social-care- policy_uk_5922bcc9e4b03b485cb2cd9b

Smith, P. (28 April 2017) The 57 Times Theresa May Has Said "Strong And Stable Leadership" So Far Since She Called The Election, Buzzfeed, https://www.buzzfeed.com/patricksmith/here-are- 57-times-theresa-may-has-said-strong-and-stable?utm_term=.eyQol76w5#.jl84LnrAk

Snelson, C. L (2016) Qualitative and Mixed Methods Social Media Research: A Review of the Literature, International Journal of Qualitative Methods, pp. 1-16, 10.1177/1609406915624574

Stromer-Galley, J. (2014) Presidential Campaigning in the Internet Age. New York:Oxford University Press.

Thelwall, M., Kappas, A. (2014). The role of sentiment in the social web. In: von Scheve, C. & Salmela, M. (eds.) Collective Emotions. Oxford: Oxford University Press, pp. 375-388

Titlow, J.P. (09 August 2017) Why Political Memes Are More Pervasive On Facebook And Twitter Than Ever, Fast Company, Available at https://www.fastcompany.com/40450279/why-facebook- and-twitter-have-struggled-to-stem-the-tide-of-political-memes

Twiplomacy, Twiplomacy Study 2017, Available at http://twiplomacy.com/blog/twiplomacy- study-2017/

Twitter Marketing (18 January 2012) Unforgettable uses Twitter to help drive tune-in during sweeps week, Twitter Blog, Available at https://blog.twitter.com/marketing/en_us/a/2012/unforgettable-uses-twitter-to-help-drive-tune- in-during-sweeps-week.html

Twitter Support (n.d.), Hashtags on Twitter, Retrieved from https://support.twitter.com/articles/ 49309

Twitter Support (n.d.), Posting photos or GIFs on Twitter, Available at https://support.twitter.com/articles/20156423#

White, A (29 May 2017) Here's What We Learned From Theresa May And Jeremy Corbyn's "Battle For Number 10" Show, Buzzfeed, Available at https://www.buzzfeed.com/alanwhite/heres-what- we-learned-from-theresa-may-and-jeremy-corbyns?utm_term=.xuJJY8kqo#.re5D09yQe

Appendix 1 - Code frame 1 RQ1

● What is the content of the memes used on Twitter in the three case studies?

Party Leader: who does the tweet and/or meme explicitly depict or mention by name? Please remember to look at both elements before putting in the code.

No Category Description Code

1 Just Theresa May You can clearly see TM's face in the meme; 1 she is mentioned in the text or the meme by name either as 'Theresa May', 'Theresa', 'May', 'Prime Minister', PM, says 'she' and in the context it is referring to TM

2 Just JC You can clearly see JC's face in the meme; he 2 is mentioned in the text or meme by name either as 'Jeremy Corbyn', 'Jeremy', 'Jezza'

3 Both TM and JC You can see both TM and JC being depicted 3 or mentioned by name in the meme

4 Other = Phrase The tweet and/or meme mentions the 4 without party phrase 'strong and stable' without leader specifically referencing a party leader (it can be related to the campaign trails, other Tory officials, the exit poll, the election as a whole etc.)

Issues: On the whole, taking into account the meme and the body text of the tweet, what issues does the tweet refer to:

1 Overall, the tweet and/or meme refers to the 1 campaign led by TM so far and the Conservative election campaign trail broadly, Election Campaign slogan, or the Conservative party and/or their HQ. It can also refer to a speech or interview TM did during her campaign. It can refer to the exit poll of the election.

2 The tweet and/or meme refers to 2 Manifesto and Conservative manifesto, or a specific policy initiated by TM & supported by the Tories policies broadly on NHS, education, elderly voters, police cuts, Brexit, the EU etc.

3 The tweet and/or meme refers to the direct 3 Competition competition between Theresa May and Jeremy Corbyn

4 The tweet and/or meme specifically mentions 4 the word leader or leadership, and it a Leadership reference to TM's leadership as a PM and as a candidate in the GE.

5 Call to The tweet and/or meme is a direct call to 5 action asking people to vote and/or openly vote/support support for TM

6 The tweet and/or meme refer to TM's 6 personality, represent a personal attack in a Personal attack ridiculing way, or using a particularly unflattering image of TM to make a point for people to vote against her

7 Unclear/Other Falls outside any of the categories above. 7

Type: What type of meme is in the tweet?

It is short animated clip, in the bottom left 1 GIF that does not corner it has the label ‘GIF’. Lasts 2-5 seconds. 1 include TM It depicts scenes from movies, shows, cartoon, series, ads.

It is short animated clip, in the bottom left 2 Manipulated GIF corner it has the label ‘GIF’. This GIF is edited 2 that includes TM to contain a loop animation of TM or JC, or their face has been imposed on the GIF.

Image of TM - no 3 3 This is a stand alone image of TM words

It is a static image (may contain 2-3 words and 4 4 Image these words are not at the forefront of the image)

The image has text on it (more than 2-3 5 words) written on top of it, make up over 75% 5 Image with text of the image, or is part of the dialogue in the photo)

Text, Articles, This is an image of a newspaper cover, a 6 6 snapshot of an article or page, simply a Newspapers segment of written text, a graph, or a

screenshot of someone's written comments.

A video that does not have the ‘GIF’ label in 7 the bottom left corner. It can depict footage that can be found on YouTube or on other 7 Video news websites.

8 Other Falls outside the categories above 8

Action: What is the action depicted in the GIF?

A person, construction, object, person + object 1 Falling/Accident is falling, breaks down, falls, part of an 1 accident or a malfunction You expect the gif to go in one direction, but it ends up being the opposite course of action. 2 Backfire 2 The author of the action in the gif has the action backfiring in their face. The characters in the gif are touching their head or their face in disbelief, they can't believe what is happening, are shocked towards what is happening, they nod their 3 Disbelief/Facepalm 3 head in disapproval, don't believe the message that is being told to them; the character can roll their eyes in disbelief, not believing what is being said The GIF depicts explosion, destruction, fire, 4 Fire/Destruction 4 flames, devastation The character is trying to hide, run escape the 5 Hiding/Dodging situation; they may also avoid a situation or 5 dodge bullets. The character expresses feeling of regret that 6 Regretting they should have not done something, or 6 something didn't go as planned 7 Fight The characters are fighting 7 The character either doesn't know what they 8 I don't know are doing, they may be floating unconsciously 8 or shrugging The character is panicking, large open eyes/mouth, or it is scared and jumps of fear; 9 Panic 9 the character could also be panicking and doing an action in a rushed manner 10 Other Falls outside of the above categories 10

Appendix 2 - Code frame 2 RQ2 ● What types of users share those memes?

User Category: Looking at information of that account, what category does the account fall into? Mainstream media, news and Media organisations, global national and 1 (mainstream and local that have both digital and non- 1 online) digital outlets. Online media: blogs, news portals This is the official account of an 2 Organisation organisation, be it a business or 2 company, account of a campaign or initiative Journalists Individuals employed by web and (mainstream non-web MSM organizations, or who 3 regularly work as freelancers for 3 media, online MSM, MS new Media, non-media media) organisations. Individuals who post regularly to an established blog, and who appear to 4 Bloggers identify as a blogger on Twitter, have 4 a BlogSpot/ WordPress link in their bio, or have the word blog in their username. Elected officials, known primarily for their membership of political parties 5 Political actors or relationship to government. This 5 includes local government, councillors. Individuals or organisations, that self-identify as an activist, or who 6 Activists appear to be tweeting purely about 6 activist topics to capture the attention of others. Individuals who are famous for reasons unrelated to technology, 7 Celebrities 7 politics or activism. Often appear as verified on Twitter. An individual who is affiliated with a 8 Researchers 8 university or think-tank. Individuals who provide no link to Members of the organization or institution. The 9 9 public account appears to be maintained by a private citizen in their personal

capacity, highlighting personal information in their bio. An account that may fit more than one of the categories 10 Unclear (blogger/activist/journalist) where it 10 is difficult to distinguish which is most important. Accounts that do not clearly fit in any 11 Other 11 category Accounts set up for the sole purpose of supporting JC, they either contain 12 JC Fan accounts his name, his hashtag, express open 12 support for him, they look like a campaign account to support him. Accounts set up for the sole purpose Mock TM of criticising or ridiculing TM, they 13 either contain her name, an aspect of 13 Accounts her appearance, or the slogan strong and stable in a ridiculing way. Political Leaning: Looking at the bio, username and photo of that user, what is the political leaning of the user?

Official This username/account is either 1 Conservative Theresa May, or the Conservative 1 Party UK party accounts/username

Official Labour This username/ account is either 2 Jeremy Corbyn himself or the Labour 2 Party UK party accounts/username The username or bio contains any of the words Labour, Socialist, Corbyn, #JC4PM, leftie, left, Green Party, Corbynist/a.

The user is on the The user works for a company that is 3 left side of the 3 associated with the left-leaning political spectrum politics such as The Guardian (workplace can be used a proxy for political leaning). They may be part of a union. They are openly dissing the government, the PM or the Tory party. The user is on the The username of bio contains any of 4 right side of the the words Conservative, Tory 4 supporter, right-wing, openly dissing political spectrum Jeremy Corbyn.

Unable to find any clues in the username of bio to categories this 5 Unclear 5 account into the two above categories.

Appendix 3 - Code frame 3 RQ3 ● What is the evaluative stance of the memes in relation to May’s political leadership?

Stance: Taking into consideration just the meme, what is its evaluative stance of the meme? It criticises Theresa May and/or an 1 Critical aspect of her campaign, leadership, 1 performance, party, policies It supports Theresa May and/or an 2 Supportive aspect of her campaign, leadership, 2 performance, party, policies It is not clear whether the tweet is in 3 Neutral 3 favour or critical Stance: Taking into consideration the body text and the meme, what is its evaluative stance of the overall tweet?

It criticises Theresa May and/or an 1 Critical aspect of her campaign, leadership, 1 performance, party, policies It supports Theresa May and/or an 2 Supportive aspect of her campaign, leadership, 2 performance, party, policies It is not clear whether the tweet is in 3 Neutral 3 favour or critical

Appendix 4 – Ethical Approval Letter

Appendix 5 – Access to Data Form Access to Dissertation

A Dissertation submitted to the University may be held by the Department (or School) within which the Dissertation was undertaken and made available for borrowing or consultation in accordance with University Regulations.

Requests for the loan of dissertations may be received from libraries in the UK and overseas. The Department may also receive requests from other organisations, as well as individuals. The conservation of the original dissertation is better assured if the Department and/or Library can fulfill such requests by sending a copy. The Department may also make your dissertation available via its web pages.

In certain cases where confidentiality of information is concerned, if either the author or the supervisor so requests, the Department will withhold the dissertation from loan or consultation for the period specified below. Where no such restriction is in force, the Department may also deposit the Dissertation in the University of Sheffield Library.

To be completed by the Author – Select (a) or (b) by placing a tick in the appropriate box

If you are willing to give permission for the Information School to make your dissertation available in these ways, please complete the following: X (a) Subject to the General Regulation on Intellectual Property, I, the author, agree to this dissertation being made immediately available through the

Department and/or University Library for consultation, and for the Department and/or Library to reproduce this dissertation in whole or part in order to supply single copies for the purpose of research or private study (b) Subject to the General Regulation on Intellectual Property, I, the author, request that this dissertation be withheld from loan, consultation or

reproduction for a period of [ ] years from the date of its submission. Subsequent to this period, I agree to this dissertation being made available through the Department and/or University Library for consultation, and for the Department and/or Library to reproduce this dissertation in whole or part in order to supply single copies for the purpose of research or private study Name GRUIA MIHAELA Department INFORMATION SCHOOL Signed MIHAELA GRUIA Date 03 09 2017

To be completed by the Supervisor – Select (a) or (b) by placing a tick in the appropriate box

(a) I, the supervisor, agree to this dissertation being made immediately available through the Department and/or University Library for loan or

consultation, subject to any special restrictions (*) agreed with external organisations as part of a collaborative project. *Special restrictions (b) I, the supervisor, request that this dissertation be withheld from loan, consultation or reproduction for a period of [ ] years from the date of its

submission. Subsequent to this period, I, agree to this dissertation being made available through the Department and/or University Library for loan or consultation, subject to any special restrictions (*) agreed with external organisations as part of a collaborative project Name Department Signed Date

THIS SHEET MUST BE SUBMITTED WITH DISSERTATIONS IN ACCORDANCE WITH DEPARTMENTAL REQUIREMENTS.