Pawpular Profiles: An Exploration on the Practice of Micro-Celebrity by Non-Human Companion Animals

(1)

1

Pawpular Profiles:

An Exploration on the Practice of Micro-Celebrity

by Non-Human Companion Animals

Daniëlle Bakker 10715584

Supervisor: Melis Baş

Second reader: Gavin Mueller MA New Media and Digital Culture University of Amsterdam

(2)

2

Abstract

Celebrity has become more mainstream with ordinary persons rising to fame. The micro-celebrity practice is one example of this development. Despite this trend, however, non-human animals have not been taken into consideration for this notion. By conceptualising micro-celebrity as a practice of the self, research has been insensitive to other actors that are part of the construction of the micro-celebrity persona. This thesis contributes to this research gap by exploring how non-human animals practise micro-celebrity. A qualitative content analysis has been carried out on six Twitter accounts that centre around non-human animals. It was found that the practice of micro-celebrity by non-human animals can be labelled in six different categories, namely ‘intimacy’, ‘ordinariness’, ‘commodification’, ‘fanbase’,

‘popularity’ and ‘for a higher cause’. In this regard, it is argued that the concept of micro-celebrity should be extended to non-human animals, for it is practised in a quite similar manner as it has been previously conceptualised for humans. However, the specific way in which it is practised suggests that micro-celebrity should not be confined to the personal aspect thereof. It should instead be conceptualised as a shared practice in which a multitude of actors, both human and non-human, play a constituent role.

(3)

3

Chapter 1: Introduction

Andy Warhol is credited with the idea that the future will see everybody be famous for 15 minutes (Papacharissi and Gibson 75). This idea suggests that there was a clear distinction between the famous and the ‘other’ in the past, but that this divide will erode. Some researchers have argued that there has indeed been a flattening of fame or celebrity culture (Turner, Celebrity, Participation and the Public 83; Hearn and Schoenhoff 203). Social media are considered to play a large role in this development (Marwick and boyd, To See and

Be Seen 139). This thesis takes up this argument in the exploration of the micro-celebrity

practice that has arisen on social media.

For this thesis, social media are understood as a ‘group of Internet-based applications that build on the ideological and technological foundations of Web 2.0 and that allow the creation and exchange of user-generated content’ (Kaplan and Haenlein 61). Ideologically, Web 2.0 supposes a participatory web in which users are both the producers of content as well as its consumers (Van Dijck, The Culture of Connectivity 4). Technologically, this has helped to give shape to a networked world in which users connect with other users online. This connectedness, as Van Dijck calls it (The Culture of Connectivity 4), goes beyond interacting with pre-existing friends.

Anyone around the world can connect with any other person who has access to social media. Social interaction is thereby quantified into data (Van Dijck, Datafication 198). This datafication, as it is called, is a paradigm that presupposes that an algorithmic identity of a user can be formed (Van Dijck, Datafication 198; Cheney-Lippold 165). An algorithmic identity goes beyond the traditional demographics such as gender or race. Instead, it is made-up of post-demographics in which interests and tastes make made-up a profile of a user (Rogers 154). This algorithmic identity is thus part of the deep personalisation that is supposed to reflect the specific interests of the user (Van Dijck and Poell 9). It can, therefore, help feed the user content that is deemed relevant for him specifically. As such, it goes against the grain of catering to a mass audience. Instead, a niche market is created (Turner, Celebrity,

Participation and the Public 91). In this long-tail logic, a new type of celebrity has arisen

online: the micro-celebrity (Marwick, You May Know Me 346).

In its first and often cited formulation by Senft, micro-celebrity: ‘is best understood as a new style of online performance that involves people “amping up” their popularity over the Web using technologies like video, blogs and social networking sites’ (Camgirls 25). Alice Marwick adds to this by arguing that it is a practice in which someone shares personal

(6)

6 information to create an online persona to be followed (Dissertation1_{13). In essence, it is} about oneself performing to an online audience as oneself.

Even though the concept of micro-celebrity is researched extensively (Hearn and Schoenhoff 2016; Khamis, Ang and Welling 2017; Marwick 2013; Marwick and boyd 2011; Raun 2017), little to no attention has been paid to non-human versions of these accounts. The celebrity status of specific non-human animals in various media has been researched (Fudge 2002; Blewitt 2013; Giles 2013), but here they have often played a role (e.g. in film or on television) or are sometimes used as the face of a greater cause in campaigns (Goodman and Littler 269). They are therefore not performing as themselves, but as something or someone else. Giles’ study of animal celebrity contains a taxonomy of animal celebrity in which he does acknowledge the existence of celebrity pets. Giles’ understands the celebrity pet as a companion animal that becomes famous by proxy, by being the pet of a famous human (119). Even though these pets may ‘play’ themselves, it is not their persona that attracts attention in the first place. As micro-celebrities, contrary to the celebrity status of non-human animals that has been researched, it is their own life that is subject of interest rather than that of a fictional character or a famous human.

Micro-celebrity, in the formulation of Senft quoted earlier, is a distinctively human practice and therefore leaves out the non-human animal rising to social media stardom. In the conceptualisation of Marwick, it is the person that reveals personal information on himself. While human micro-celebrities present themselves or ‘type oneself into being’ (Sunden qtd. in Rogers 155), non-human animals cannot be said to have this same authorship over their representation. Even though it may look as if they are the ones to do the talking, the

representation is co-crafted by a human who has some relationship with the non-human. This co-crafting raises some severe questions since the idea of micro-celebrity is showing one’s authentic self, creating one’s (online) identity and personal life (Marwick and boyd, To See

and Be Seen 146) to be shared with others.

It is this relatively uncharted territory that this thesis wants to explore. It takes an interdisciplinary approach as it draws from theories and methodologies of Media Studies and both theories of Celebrity Studies and Human-Animal Studies. The research question that this thesis explores is: How is the practice of micro-celebrity constructed and performed on

non-human companion animal accounts on Twitter?

1_{I have used both the dissertation version of Status Update as well as the book version. To make this distinction}

(7)

7 In order to answer this question, both a theoretical as well as an empirical approach were taken. The introduction is written around the central theme of micro-celebrity. It takes up questions of authenticity, agency, and privacy that arise when this concept would be applied to non-human animals. It answers the sub-question: What makes the concept of

micro-celebrity possible for non-humans to practise?

The second chapter, the literature review, explains the concept of micro-celebrity further. The chapter draws on multiple conceptualisations of diverse scholars to show how the micro-celebrity practice has been defined in human terms as a personal practice. The underlying argument is that the concept should be redefined to account for non-human animals that also practise micro-celebrity. It answers the sub-question: How has the concept

of micro-celebrity been conceptualised in previous research?

In chapter 3, the methodology, an in-depth outline is given on the steps that were taken to complete the case study of chapter 4. It answers the sub-question: How can the

practice of micro-celebrity by non-human animals be analysed?

In chapter 4, a qualitative content analysis using both emergent coding and a priori coding is carried out on six non-human animal accounts on Twitter. This analysis mainly focuses on how these accounts are practitioners of micro-celebrity. It answers the sub-question: How is micro-celebrity practised by six non-human animal Twitter accounts?

In chapter 5, the case is made that micro-celebrity is indeed in need of an extension to include a multitude of actors that play an integral role in the practice, both human and non-human. It answers the sub-question: How should the practice of micro-celebrity be

conceptualised to account for non-humans? Furthermore, it considers possible opportunities

in presenting a non-human animal as micro-celebrity. It answers the sub-question: What does

this conceptualisation mean for how humans perceive of non-human animals?

Lastly, chapter 6 provides a small overview of all the conclusions made earlier in the thesis and suggests possible avenues for further research.

1.1 Non-Human Persona

A micro-celebrity is a persona that aims to gain a following, fame or even revenue by

performing as oneself. However, in the case of non-humans, the online persona is co-created for them by a human. For non-humans to be regarded as micro-celebrities, then, they should be considered to have a persona.

According to the Cambridge Dictionary, a persona is ‘the particular type of character that a person seems to have and that is often different from their real or private character’

(8)

8 (dictionary.cambridge.org). It is thus understood as how a person is perceived to be, rather than the exact way one really is.

This definition opens up the idea that a non-human animal can be a person and indeed have a persona. A human can then write about the personality of a non-human animal, as a persona can be at odds with how the creature itself experiences its character to be. Indeed, psychologist Gosling even argued that judging an animal’s personality may be best done by others (read: humans) who know them, for they cannot speak to us in the same manner as humans can (1161). If a dog looks as if he/she is comforting his owner who is in distress, it may be interpreted as such. This anthropomorphising, the interpreting of a non-human as having shared qualities with humans, makes it the ideal candidate to have a persona.

This argument thus presupposes that a non-human animal can have a personality: different character traits that combine into a perceived single characterisation of how a person is. This idea is not wholly controversial in contemporary Western thought. It is for example expressed in but not limited to poetry (Hollander 470), some childhood books in which non-humans star (Fudge 73) and (animated) movies and television series (Fudge 80). Companion animals, in particular, as individualised non-humans, as both animal and human at once (Fudge 28), seem to have this claim to personality. As beings that are close to humans, as their actions can be seen and felt, they can be interpreted in a particular manner (Fudge 33).

In the interpretation of how a companion animal is, and the dissemination thereof online, there is thus a serious role to be played by the human who writes the persona into being. The non-human animal, however, has some agency on the way he or she is

represented. Agency is here understood in a Latourian fashion as the ability to act or have an effect on the world (Latour 71). It presupposes that agency is not something to be ascribed to humans alone or only beings that can act intentionally. It, therefore, goes against the

definition by Ortner that defines agency as intentional action (Ahearn 35). Similarly, it defies Kockelman’s description of agency as ‘the relatively flexible wielding of means toward ends’ (Kockelman qtd. in Ahearn 36).

Instead, Latour’s conceptualisation holds that non-humans do possess the ability to alter the world with their actions. The non-humans should, therefore, be considered to be agents. This includes, among others, both non-human animals and software. Adding to this, Donna Haraway argues that agency is almost always unequally distributed among different actors within a power relation (262-263). In this specific instance, the human as the writer of the non-human animal accounts is in a powerful position to represent the lives and possible

(9)

9 thoughts of non-humans online. It is this unequal power relationship in the creation of the micro-celebrity that merits examination.

It is even more crucial to research this phenomenon, for the practice is relatively easy to replicate. This ease may subject other non-human animals to the practice as well. Micro-celebrity does not necessarily require a tremendous amount of wealth, nor technical skills or a hard to come by distribution platform because it is done through social media. The follower of said persona may be aspiring to practise micro-celebrity with their ‘own’ non-human animal. He can learn from other representations to craft his ‘own’ to enter the field as a content creator and potentially influential actor himself (Marwick, Status Update 14-15). As such, past representations of non-humans as micro-celebrity may lead to future

representations of other non-humans in a similar manner.

1.2 Non-Human Micro-Celebrity and the Paradox of Authenticity

Visibility and attention are both the prerequisites and the aims of micro-celebrity (Khamis, Ang and Welling 198). For micro-celebrities to grab and sustain attention, there seems to be a paradox in place. It is about an ordinary person living an ordinary life. At the same time, this life should be eventful enough to be shared and followed. This friction potentially creates a problem. How should one ‘brand’ one’s life to fulfil both the ordinariness and the

extraordinary part of the equation? If someone’s life is not deemed attractive, it is unlikely that the person is followed. At the same time, it is this ordinariness that attracts. It is this tension that may lead to a problem of authenticity. Marwick has argued that authenticity is a significant factor in the micro-celebrity practice (Status Update 119). One must keep close to oneself, being ordinary and at the same time, stand out enough to be recognised as worthy of being followed.

Authenticity is here defined as the act of expressing one’s true self, thus being genuine to one’s personality – at least in the eye of the beholder. Authenticity is a judgment made by others on the performance of a persona (Marwick and boyd, I Tweet Honestly 124; Thomas 253). Following the line of thought set out in the preceding paragraph, there is already difficulty in being authentic for humans performing as themselves. However, with non-humans posing as themselves with the intervention of the human, this becomes even more problematic.

For David Giles, non-human animals can easily be said to be ‘authentic’ (117). They do not have ulterior motives such as making a profit or wanting to be famous so badly, consequently altering their behaviour. Therefore, he argues, they are always seen as ‘being

(10)

10 themselves’ (Giles 117). However, this is only true for him once the non-human animals are stripped from ‘all human pretence’ (Giles 117). The non-human micro-celebrities are

subjected to the motives, ideas, and beliefs of their human companions. Therefore, it becomes hard to make that same claim to authenticity. Indeed, Marwick and boyd seem to believe that if another person writes the content than the one the social media account is about, it cannot be ‘authentic’ (To See and Be Seen 142-143). This line of thought begs the question if authenticity should even be considered a part of the micro-celebrity practice of non-human animals to begin with.

If a human is already incapable of reading the mind of another person, or interpreting another man’s experiences in the same way, even through language, how is it possible for a human to tap into the experience of another animal? In this regard, authenticity seems to be hard to achieve. In chapter 4, it is explored whether authenticity remains a key characteristic of micro-celebrity when non-humans practise it, and alternatively what other characteristics of the practice can be found.

1.3 On Private, Public and the Self

As should be clear by now, even though the life of the non-human animal is shared to create and sustain an audience, it is not the non-human who is in charge of the release of the

content. Apart from the complicated dynamics already touched upon in the previous sections, there is an even deeper issue to be discussed: the apparent conception that one is allowed to share the life of non-human animals online. In this regard, it seems to be strongly connected to the notion of privacy. The term ‘privacy’ has many definitions, stretching from ‘the right to be left alone’ to information control to autonomy (Moore 215). In this thesis, privacy is the right to have a certain level of control over sharing one’s personal life with others (Moore 218). This definition suggests that a certain level of agency is needed on the part of the one whose life is being shared while suggesting that this may not always be the case if privacy is infringed upon. This definition is, therefore, fitting to use when talking about non-human micro-celebrity.

Media scholar Brett Mills contends that privacy seems to be a right that is hardly if ever bestowed on non-human animals (196). Following Tobias Raun, micro-celebrity as a practice can be thought of as a genre of intimacy (105), which requires private life to be negotiated and served to a public. Sharing the life of one’s companion animal on social media is an obvious way of letting others into that private life and suggests that the human

(11)

11 deemed fit for online sharing. Even though human personas struggle to find a proper division between what should remain private and what can be shared (Baym 25), the same privilege does not seem to apply to non-human animals. They are instead constantly subjected to a human gaze (Malamud, Animals on Film 6-7), for if the non-human animal does something worthy enough of being posted, it should be captured to be shared. They are arguably the most fitting to become this kind of persona since they are often close to humans, both emotionally and physically. Indeed, just as Rembrandt van Rijn made many self-portraits because he was a model that was always available (Rothenberg 109), so too is the companion animal.

Concern over privacy keeps growing, especially in the digital environment in which ever newer data is generated on various aspects of society (Van Dijck, Datafication 197). Paradoxically, anxiety over the protection of privacy of non-human animals lacks (Mills 196) even though non-human animals do require it in a way (Moore 215). New technologies make it possible to keep constant watch over them or track them. These practices stretch from wildlife to zoo animals to companion animals and are employed with a variety of purposes. These include but are not limited to: camera traps hidden from sight to film or photograph wildlife; drones equipped with cameras that can soar high to film animals; crittercams attached to animals to track their movement; ZooCams that act like webcams to film zoo animals; GPS technologies to track a variety of animals; the chipping of animals such as cats or dogs and live streams of companion animals. To research all these phenomena is far beyond the scope of this thesis. It is, however, relevant to note that the lack of consideration for the privacy of non-human animals is not confined to the practice of micro-celebrity but is thus part of a larger development for it is one of many practices that infringe on the privacy of non-human animals.

At the same time, humans become increasingly wary of their own rights to privacy (Brandtzæg, Lüders and Skjetne 1006-1007). If humans themselves are not ready to be subjected to this kind of visibility, why would it be permissible to do this to non-human animals? The possibility of capturing and sharing the lives of non-human animals because they cannot object in the same manner as humans can, does not mean that they comply nor that humans are allowed to take advantage of this position.

Though it may not necessarily hurt the companion animal to have his life shared online, in the words of Randy Malamud: ‘is that the best we can hope for?’ (Animals on Film 12). Malamud argues that not doing any harm to a non-human animal should not be enough to decide to use the animal in a particular way. Instead, he argues, that we should strive for a

(12)

12 representation of non-human animals in our media sphere that can actually help them

(Malamud, Animals on Film 18). I will return to this point in later chapters.

The non-human animal cannot be said to choose to be represented in the first place, let alone how. Therefore, I argue there is an unequal power relation at work here. From a Human-Animal Studies perspective, it is contended that non-humans do have a form of agency on their representations (Smaill 154; Burt 31). A relationship only comes into being as two or more actors relate to each other. It is, therefore, always a process of social co-shaping (Pearson 244) rather than one party exerting all power and total control. That said, the human does have a significant role in this process and a lot to gain from the practice of micro-celebrity. As noted earlier, it follows a logic of commodification. The non-human animal and his life become a commodity which can be sold.

1.4 Commodifying A Persona

Self-branding, a concept linked to micro-celebrity, is described by Khamis, Ang and Welling as a practice in which an individual crafts a persona for financial benefit or cultural capital (191). However, as Marwick notes, even though these are two separate terms, they do have similarities. The celebrity, she argues, is a ‘person-as-commodity’ (Marwick Dissertation17). The celebrity sells his persona to a public.

As social media accounts of practitioners of micro-celebrity are made to be seen and to become famous in the process, the personas may become what is called ‘influencers’. An influencer is here understood as someone who converts the (micro-)celebrity status into value by creating a personal brand that appeals to either commercial or non-commercial outlets which can use this persona for advertising their goods or cause to a public (Hearn and Schoenhoff 194). Companies and the like can thus capitalise on the celebrity status of a persona to influence consumer decisions. This technique echoes word-of-mouth rather than a straight-up endorsement by a celebrity in advertising, which is thought to be more effective (Hearn and Schoenhoff 203). Apart from influencing consumer choices, these representations by the sheer volume of people they reach, can influence how users may be inclined to think about the persona in particular and non-human animals in general. This point is touched upon further in chapter 5.

Micro-celebrity is thus intricately linked to commodification. The non-human animal is a commercial opportunity for any aspiring (human) user who wants to run a successful micro-celebrity account without starring in it himself. As is widely known, ‘cute’ sells (Meese n. pag.). Non-human micro-celebrity is the manifestation of this principle in a

(13)

semi-13 amateur way. The ease with which a non-human animal, especially a companion animal, can be photographed or told stories about is, I would argue, one of the reasons why this type of micro-celebrity account is so prominent. This is coupled with the reasons mentioned earlier: the lack of an idea of privacy for non-human animals; the lack of technical skills needed or financial investment necessary to practise it. Another crucial factor in this regard is the accessibility of a social medium that works as a distribution platform.

1.5 Software as Actor

For a persona to practise micro-celebrity, both visibility and attention of others are needed. (Khamis, Ang and Welling 198). As social media provide this, the practice cannot be separated from the social media that facilitate it. In the case study of this thesis, Twitter was chosen as a particular social medium that enables micro-celebrity. Twitter, as Marwick and boyd have shown, is a key social medium for self-branding and micro-celebrity practices (Marwick and boyd, To See and Be Seen 144; Marwick and boyd, I Tweet Honestly 116). It offers some affordances that are key to the practice of micro-celebrity. This will be further explained in chapter 3.

A social medium such as Twitter shapes the process of meaning-making. A social medium is an infrastructure that ‘set[s] the stage’ (Bratton 47) for the interaction with that platform by the user, as it affords certain kinds of actions while making others impossible or less likely (Davis and Chouinard 244). It thus has a part to play on the level of content as well (Gillespie 358). A social media platform is thus not a mere in-between, but an actor that is part and parcel of the process.

At the most basic level, in the case of Twitter, it only affords tweets that are a maximum of 280 characters long. The content of one tweet is, therefore, ought to be condensed. Furthermore, Twitter asks its users ‘what’s happening?’, which suggests that a user answers this question with his tweet rather than providing other content. It encourages apart from the input of text and emojis to add photos or videos, GIFs, a poll or one’s location. A content type that is missing from this, for example, is sound (MP3/MP4). In this short analysis of Twitter’s affordances, it already becomes clear how it is a co-shaper of content. I argue that it is essential to see how Twitter is an actor in shaping the micro-celebrity practice.

Following Gillespie’s influential work on YouTube, Twitter as a platform serves a multitude of actors with different needs simultaneously (353). It caters to end-users,

advertisers and marketers, governments and itself. Researchers are, however, not its core user type to cater to but its ambiguous position as a platform does shape the data that is researched

(14)

14 in this thesis. Twitter data are thus not created with the purpose to be researched. To account for this, it is opted here to obtain the data for this thesis via a toolset that is created with the intent to conduct research with (Borra and Rieder 266). The DMI Twitter Capture and

Analysis Toolset (DMI-TCAT) works through Twitter’s Application Programming Interfaces (APIs). These APIs allow third party software programs, such as DMI-TCAT, to access features of Twitter (Borra and Rieder 267).

The developers of DMI-TCAT rightly contend that the tool shapes, just as Twitter does, which kind of research can be undertaken (Borra and Rieder 269). This influence is, among other reasons, due to it following Twitter’s structure and accessing and acquiring the data through Twitter’s APIs (Borra and Rieder 267). DMI-TCAT is not just a tool for data acquisition but has built-in options for analysis as well (Borra and Rieder 269). It enables qualitative content analysis, among others. This analysis type was conducted in this thesis. Here, it becomes apparent once more how software, in general, is an actor in the research process.

Following DMI-TCAT’s affordance, a qualitative content analysis that uses both a priori and emergent coding was conducted on various tweets of Twitter accounts in which companion animals star. As suggested earlier, the proliferation of pet accounts on social media, that I argue should be thought of as practising micro-celebrity, is not entirely coincidental. As beings that are both physically and emotionally close to humans, the experiences of their lives are particularly suited to be shared by humans online. For the current research, a total number of six accounts were analysed. Even though the analysis was carried out on different Twitter accounts, I argue that the data sets that were extracted are fairly commensurable. Twitter, as is explained in greater detail in chapter 3, already forms and informs what kind of messages can be sent. The exact steps that were taken are detailed in the methodology section, chapter 3 of this thesis.

1.6 Overview

The following chapter contains a review of relevant literature that was already briefly touched upon here. It provides a more thorough theoretical understanding of micro-celebrity as a personal practice. In chapter 3, the methodology of this thesis is explained in more detail. It describes the steps that were taken to conduct the case study of the thesis, chapter 4. In that chapter, a qualitative content analysis has been carried out. Chapter 5 elaborates on what the analysis means for the concept of micro-celebrity, furthered by discussing what this new conceptualisation may mean for how humans think of non-human animals. Finally, chapter 6

(15)

15 provides a summary and ideas that can be pursued in future research. These chapters together try to answer the research question: How is the practice of micro-celebrity constructed and

(16)

16

Chapter 2 – Literature Review

This chapter reviews existing literature on the concept of micro-celebrity. It provides context by identifying the scope of the current knowledge on the concept, problematises the current interpretation of its context and establishes a research gap that this thesis tries to fill. It argues for a redefinition of the concept of micro-celebrity and related concepts to include

non-human animals. It answers the sub-question: How has the concept of micro-celebrity been

conceptualised in previous research?

2.1 Micro-Celebrity: A Personal Practice

The concept of micro-celebrity was first coined by Theresa Senft in 2001 and published in her 2008 piece: Camgirls: Celebrity and Community in the Age of Social Networks. Senft studied camgirls, women who try to become famous by broadcasting their lives online via webcams (Camgirls 1). Even though the camgirl practice had already seen its heyday by the time Senft published her book (Camgirls 11), the idea behind the sharing of personal lives with an online audience is far from old. Digital technologies such as social media have made it easier for anyone with internet access to share personal information online with a large user base and possibly rise to fame in the process (Marwick, Status Update 115). Unsurprisingly then, her concept of micro-celebrity has been taken up and is still researched extensively.

Curiously, Senft’s first own definition lacks the idea that micro-celebrity is a personal practice, or at least concerned with the personal lives of the practitioner. Micro-celebrity, in Senft’s words, is ‘a new style of online performance that involves people “amping up” their popularity over the Web using technologies like video, blogs and social networking sites’ (Camgirls 25). The ‘performance’ she spoke of does not necessarily seem to be a

performance of the self, as this is lacking in her conceptualisation.

In her later work, however, Senft sees micro-celebrity as ‘a new form of identity linked almost exclusively with the internet’ (Microcelebrity and the Branded Self 1). Here, the concept does not just entail anyone performing whichever role possible, but it is tied to an identity. Micro-celebrity, it seems, has become a personal affair as can be interpreted from Senft’s above-quoted words. Indeed, it is precisely this addition to the original concept that seems to have taken over. This addition proves problematic for non-human animals who are practitioners of micro-celebrity. They are not the ones who construct the identity, but their lives are at the centre of the practice.

(17)

17 Alice Marwick and danah boyd conceptualised micro-celebrity in their article To See

and Be Seen: Celebrity Practice on Twitter as follows: ‘[It] can be understood as a mindset

and set of practices in which [an] audience is viewed as a fan base; popularity is maintained through ongoing fan management; and self-presentation is carefully constructed to be consumed by others’ (140). It is thus a presentation of the self rather than an undefined performance or a fictional role to be played out. It is someone performing as a version of oneself.

In her book Status Update, Alice Marwick takes this notion a step further. Rather than speaking of a broad concept of micro-celebrity, she creates a distinction between ascribed and achieved micro-celebrity (116-117). These are not two mutually exclusive categories, but they are two sides on a scale. An ascribed micro-celebrity is a person who is treated by others as a celebrity, whereby he is applauded for his accomplishments (Marwick, Status Update 116). The achieved micro-celebrity comes closer to what has been talked about in this thesis up until now. Marwick explains that achieved micro-celebrity is a kind of self-presentation strategy. It is this notion of achieved micro-celebrity that takes the personal in the concept of micro-celebrity to be a central component. An online persona is created by sharing personal information with an audience that is treated as a fanbase with which an interpersonal

relationship is constructed and maintained (117). This definition suggests even more than Senft’s later conceptualisation that the practitioner of micro-celebrity creates a highly personal, intimate identity.

This possibility of self-identity formation is evidenced further in another related concept, the DIY celebrity (Turner, Celebrity, Participation and the Public 86). Graeme Turner conceptualised micro-celebrity as a form of Do-It-Yourself celebrity that uses Web-based media as its primary outlet (Understanding Celebrity 71). He ascribes this notion to John Hartley. The latter spoke of the DIY citizen, which is different from the DIY celebrity. Hartley’s Do-It-Yourself citizen recognises the potential to choose and create a never fixed identity for oneself (Hartley 178). This identity-formation is aided by the opportunities provided by the media sphere – in the broadest sense possible – in which ordinary persons become the subject of media content. This flexibility can then lead to persons creating a persona to become famous, the DIY celebrity.

As the concept of micro-celebrity has been altered to become a personalised practice of identity formation, it appears that the constructor of the identity is necessarily the same as the persona that is created. It is this leap of logic that this thesis tries to overthrow to account

(18)

18 for non-human micro-celebrities, as they are not the sole constructors of the persona but are represented as such.

2.2 Ordinary Persons Rising to Fame

As stated earlier, the idea of micro-celebrity as a practice is that it is not confined to the traditionally famous but is open to ‘ordinary’ persons as well. Even though this widening of opportunities for the ‘ordinary’ person is acknowledged by many (Turner, Celebrity,

Participation and the Public 86), as will be explained in this section, it seems that

non-humans are still excluded.

In their article To See and Be Seen, Marwick and boyd argue that famous people, who have become famous in another way than through social media, are mimicking the practices which are conceptualised as micro-celebrity to sustain an audience. Celebrity, they argue, is not the antonym of ordinary anymore, a hard-dividing line between the famous and the non-famous. Instead, it exists on a continuum in which multiple persons with various levels of fame deploy similar strategies and practices to present themselves as celebrities (Marwick and boyd, To See and Be Seen 140). This development shows how celebrity is becoming more mainstream.

Along these lines, Oliver Driessens argues that a larger process of democratisation of celebrity takes place. He develops his argument through a historical analysis of ‘celebrity’. For Driessens, the idea of celebrity has shifted from something that is defined by a

predetermined hierarchical position, e.g. by being a noble or a religious leader (373) to a system in which accomplishments of the individual matter rather than descent (374). Afterwards, celebrity became professionalised through the use of strategic communication and promotion (Driessens 375), the Hollywood star system is a primary example of this. The last shift Driessens sees, is the increase of opportunities for ‘ordinary’ persons to become micro-celebrities, what he calls the ‘democratisation of visibility’ (380). It seems that with every new era, celebrity becomes more open to a variety of people.

Graeme Turner agrees that ‘ordinary’ persons are increasingly visible in the media sphere. He coins the term ‘demotic turn’ for this shift (Turner, Celebrity, Participation and

the Public 83). Turner, however, chose the word ‘demotic’, which means ‘of or for the

common people’ (Ordinary People 1), rather than democratic as Driessens conceptualised it. In Turner’s view, some researchers, such as Hartley and Driessens who were mentioned previously, are too quick to equate the broadening of opportunities for ‘ordinary’ persons to become visible and famous to the democratic effect this may have (Turner, Celebrity,

(19)

19

Participation and the Public 88). Indeed, as is argued here, there are various reasons to be

sceptical of this alleged democratic appeal.

Celebrity still necessitates a hierarchical position (Turner, Understanding Celebrity 93). Drawing on Bonner and Couldry, Turner argues that even the ‘ordinary’ has to stand out in one way or another (Turner, Understanding Celebrity 89-90). Indeed, ordinariness seems to be coupled with some form of extraordinariness in the pursuit of micro-celebrity.

Marwick also makes this hierarchical need clear in her formulation of the ascribed micro-celebrity. One is regarded as a micro-celebrity by others by virtue of his

accomplishments, which set him aside from the rest. Indeed, Marwick and boyd add to this by stating that the fan-celebrity relationship is always one of unequal status, even though Twitter may seem to make this distinction less explicit by providing a way for fans to interact with celebrities (To See and Be Seen 155-156). In this regard, it does not seem fitting to think of ordinary persons rising to fame as a democratic opportunity.

2.3 Social Media as Actors in the Micro-Celebrity Practice

Social media, and Twitter here specifically, are actors that shape the practice of micro-celebrity. Turner argues that the micro-celebrity acts in an environment that is increasingly structured as and, integrated into, a media conglomerate (Understanding Celebrity 93). Far from a horizontal, bottom-up, grassroots environment, social media such as Twitter and Facebook are driven by large, commercially motivated companies that do not necessarily cater to the needs of individual users but instead to a multitude of actors, including themselves (Turner, The Mass Production of Celebrity 157; Gillespie 353). Hearn and

Schoenhoff agree and add to this by stating that micro-celebrities on social media do not have ownership rights on nor control over their public persona (208). The actual monetary value that is generated in the process is likely to go to the social medium. The micro-celebrity will not earn money if this person does not capitalise on his success by being paid (Hearn and Schoenhoff 208; Marwick and boyd, To See and Be Seen 155), for example by becoming an influencer.

The dominant position of social media stretches even to content creation. Khamis, Ang and Welling in their article Self-Branding, ‘Micro-Celebrity’ and the Rise of Social

Media Influencers note that social media have inherent metrics that show one’s success (197),

what Marwick calls ‘status affordances’ (Status Update 75). These can indicate if one’s content is successful. A higher metric, for example, more followers, more likes or more shares, signifies greater success. The micro-celebrity can analyse these metrics and alter the

(20)

20 following content accordingly for maximum interaction. Khamis, Ang and Welling argue that visibility and attention are the two factors driving micro-celebrity (198), which makes it more probable that content will be tailored to achieve this. Indeed, Marwick even goes so far as to argue that social media with their status affordances function as Foucauldian ‘technologies of the self’ (Status Update 14). This term is used to describe how a technology can produce a subject that can transform the actions, thoughts, and feelings of an individual with a specific aim for improvement (Foucault 18). The central claim Marwick makes in her book Status

Update is that social media teach how to present and promote oneself, how one is deemed to

be successful, as they reward behaviour that fosters interaction and visibility (14). This argument shows how social media are integral parts of the micro-celebrity practice.

It is important to note here that it may not actually be true that higher status metrics instantly allow for higher visibility. A social medium makes use of multiple algorithms that rely on a variety of factors to curate the visibility of content. The exact workings of those algorithms are unavailable for the researcher or any other end user to explore. An algorithm works as a black box: it is provided with input, and it gives back an output. What happens in-between often remains a mystery (Bucher, Want to be on the Top 1176). However, this does not mean that a user does not have any idea of how the algorithm would work or cannot act on this supposed knowledge. Bucher provides a suitable concept to account for this

phenomenon: the algorithmic imaginary (The Algorithmic Imaginary 39-40). In her article, Bucher shows that some users alter their own media practices on their understanding of how an algorithm works (The Algorithmic Imaginary 40). Thus, even though Twitter as a social medium does not dictate what content a user must provide, how it is assumed to work can shape the user practice, nonetheless.

2.4 Micro-Celebrity: Not Democratic nor Demotic

The claims made by the various authors referenced in the previous section show that even though the status of celebrity has arguably become more readily available, one cannot safely assume this means the playing field has become more democratic. As shown, there are still constraints in place that limit the power of the micro-celebrity to be genuinely free to craft one’s identity as one pleases.

In line with Turner’s view, I argue that one should be wary of claiming democratic potential. As already claimed in chapter 1, social media ‘set the stage’ (Bratton 47) for the micro-celebrity practice and therefore actively shape it. The democratic ideal almost suggests a technologically deterministic view that the use of technology, such as a social medium,

(21)

21 necessarily brings forth an increase in power. Power is always distributed and unequal

(Haraway 262-263) and should not be thought of so lightly. There are always various factors and actors in place in a power relation (Pearson 244).

While agreeing with Turner’s wariness on the democratic potential of ordinary persons rising to fame (Celebrity, Participation and the Public 88), I disagree with Turner’s conceptualisation of this development as demotic. As said, demotic means ‘of or for the common people’ (Ordinary People 1). It suggests that a human can solely achieve the status of (micro-)celebrity. As becomes apparent in the case study, and already touched upon multiple times until now, non-humans can be practitioners of micro-celebrity as well. It is not solely a human affair. It is, therefore, necessary to revisit the concept of micro-celebrity and the idea behind the demotic turn to account for non-humans as well.

(22)

22

Chapter 3 – Methodology

This chapter outlines the methodology that was used in the case study. It explains which steps were undertaken and what reasons lie behind the decision to opt for these. Firstly, it explains why Twitter was chosen for this project. Afterwards, a general discussion on the usage of Twitter data for research is given. This is followed by the specific way in which the data for this thesis has been acquired. This chapter concludes with a detailed roadmap for the

qualitative content analysis from chapter 4. It answers the sub-question: How can the practice

of micro-celebrity by non-human animals be analysed?

3.1 Choosing Twitter

Twitter is a social medium that affords its users in a 280 characters maximum post to share so-called tweets. These are then disseminated into the content feed of a user that follows the account from which the tweet was sent. Additionally, the tweets can be found by other users on the profile of that account, if it is a publicly available account, without the need to follow the account in question. Twitter is a ‘broadcast medium’, which makes making accounts private senseless (Marwick, Status Update 101). Thus, a user writes something for it to be seen by others. This is even more apparent in the practice of micro-celebrity. As has become clear by now, visibility is aimed for and can be considered a requirement of the practice.

It was chosen to opt for Twitter for the case study rather than another social medium for several reasons. Firstly, the content on Twitter has a specific emphasis on words, at least more than Instagram and YouTube, which emphasise visual content. Twitter, therefore, deals with a particular kind of representation. It can look as if the companion animal is uttering the words, ascribing the content that is shared from that account to him. In a way, it parallels animated movies in which animals ‘talk’ with a human voice. There is, of course, one clear distinction, the movie animals are animated, whereas the ones in the Twitter account are physical beings. Twitter is a good starting point for an analysis that deals with the (lack of) agency of companion animals to be presented online and in what way.

A social medium that shares this emphasis on words with Twitter is Facebook. There is, however, a difference in the structures of the two social media that makes Twitter more suitable for the practice of micro-celebrity. Twitter’s structure is arguably more open than Facebook’s. The latter makes two-way connections between ‘friends’. If one befriends someone else, the other becomes friends with the befriender as well. The content of personal

(23)

23 accounts is because of this structure not necessarily available. Facebook Pages are more open than personal accounts, as their content is public. Facebook’s guidelines state that Pages should only be used by businesses, organisations or for ideas (Facebook.com). A Page rather than a profile already signifies branding. It, therefore, goes against the grain of

micro-celebrity, since this can interfere with the claim to authenticity where a person shares a personal life with others.

On Twitter, on the other hand, potentially anyone can view another account’s tweets if that account is public, which is the default setting. As Marwick and boyd rightly contend, the follower/followed relationship is technologically distinct (To See and Be Seen 142). One can follow an account, without that account having to follow back to be able to see its content. This structure makes it possible to have a vast number of followers, without having to return the courtesy. This more open structure than on Facebook makes Twitter a social medium that better facilitates the practice of micro-celebrity and therefore suits this thesis better.

Furthermore, Twitter does not have a real name policy, whereas Facebook does. Twitter’s guidelines permit the usage of other usernames that are different from the name the person holds in daily life (Twitter.com), whereas on Facebook profiles need to have the same name as the one on one’s ID (Facebook.com). This policy makes Facebook arguably less suited for the maintenance of non-human animal accounts by humans than Twitter.

3.2 Using Twitter Data for Research

Twitter, as a platform, caters to the desires of a multitude of actors simultaneously (Gillespie 353), though not necessarily to the needs of academia. However, as stated earlier, this dynamic does shape the data for research.

Since the data are not explicitly intended for research, there are some questions to be addressed. Davies argues that we are constantly subjected to all kinds of tests and are ‘living in the lab’ (200; 208). That said, the users of which the data is collected do not know, nor could have reasonably expected, that this data would be used for this type of research.

Therefore, it is deniable that these users have given informed consent to be part of this thesis. In an interesting turn of events, just as the non-human animals are unable to know that their lives would be broadcasted online, now the humans who have made the content and run the account are also subjected to scrutiny beyond their reasonable expectation.

This unconscious participation in a study can be beneficiary to come close to an ideal of ‘truthfulness’ of data, Davies contends (208). The use of social media data would be more

(24)

24 reliable than the use of surveys, as unconscious behaviour would provide more truthful data than does conscious behaviour. However, this can be a false belief in the truthfulness of data (Davies 207-208) just because it was unconsciously acquired. Even though the data were not created for research purposes, they could have been made with ulterior motives in mind that can also decrease trustworthiness. In the case of micro-celebrities, this could have been to direct attention towards the account, sustain an audience, create monetary value or something else entirely. Thus, rather than seeking the ‘truth’ or treating the data as ‘truthful’, it is the way it can be interpreted that is of interest in the current study. Indeed, the data that is used here is not randomly selected to speak to some truth about social life. Instead, it was carefully chosen to fit within a study into the practice of micro-celebrity of non-human animals.

3.3 Sampling

As stated earlier, a qualitative content analysis has been conducted in chapter 4 on companion animal accounts that, as is argued here, follow the logic of micro-celebrity. The first step in acquiring the data for the analysis was to choose which accounts to use. Since this is a qualitative analysis, I chose to opt for a relatively small corpus of six accounts in total. The scope of this corpus is in part motivated by a manual analysis of all tweets in the data set to use emergent coding on.

For building a list of Twitter accounts, a selection was made of multiple possible accounts. To select the corpus for this thesis, a Google Search was first conducted for the terms “Twitter AND animal accounts” and “Twitter AND pet accounts”. These queries gave back a list of results with listicles of top Twitter accounts in which animals or specifically companion animals star. The accounts listed in the articles were looked up to see if they suit the research purposes of the current thesis. This list-building technique used an editorial approach. It made use of readily available lists which others had compiled to create a new list. The corpus of this thesis is derived from this new list. Sampling is at the heart of the research process as it informs the kinds of questions one can ask up until the results of said research. A researcher should base the sample on carefully made decisions (Rapley 49). For this reason, the sampling process is provided in detail here.

A first shift was made between accounts from or starring companion animals and those that feature non-human animals in another relationship with humans. As already explained earlier, companion animals are the most accessible non-human animals to make a Twitter account on, as they live with humans and often form intimate relationships with them. A second factor that was taken into consideration to select suitable accounts is the number of

(25)

25 followers. The follower count is a particularly visible ‘status affordance’ of Twitter that can signify the reach of content disseminated by the account (Marwick, Status Update 75). As visibility is of utmost importance to the practice of micro-celebrity (Khamis, Ang and Welling 198), the number of followers was a good indicator to use. Thirdly, the account had to have one companion animal that is written about. A micro-celebrity is centred around a persona, which is one entity rather than a multitude of individuals. Lastly, only accounts were chosen of which the companion animal could be said to have garnered attention for being him or her, rather than being famous by proxy. This, however, does not mean that there cannot be any affiliation with a more or less famous human, but this should not be the core reason for the micro-celebrity status. While taking the aforementioned factors into account, a final list was made of six accounts in total. It is comprised of three cats and three dogs.

The following accounts were analysed in chapter 4: Jiffpom (@jiffpom)

Doug (@itsdougthepug) Manny (@FrenchieManny) CatFoodBreath (@Catfoodbreath) Lil’ Bub (@IAMLILBUB)

Sockington (@sockington)

In one instance, the human companion of the cat Sockington was indeed someone famous, namely historian/filmmaker/archivist Jason Scott. However, the cat had at the time of writing 1.2 million followers, whereas Jason Scott’s account @textfiles had 27,800 followers. This difference led to the conclusion that the fame of Jason Scott is not the deciding factor in the fame of Sockington.

A notable account that was left out of this list is @RealGrumpyCat. Even though it arguably ticked all the boxes, a scroll through the posts revealed that the account is driven mostly by retweets rather than utterances of the account itself. Furthermore, many posts that did not originate from retweets do not have any words in it, only pictures. This account, therefore, did not seem to correspond to the ‘device culture’ (Weltevrede and Borra 2) of Twitter, which was outlined earlier in this chapter. @RealGrumpycat posed an analytical problem for this thesis since it often lacks the use of words. It was, therefore, excluded from this analysis.

(26)

26

3.4 Acquiring the Data

After the accounts were selected, the tweets had to be retrieved to obtain a data set to analyse. Even though, as already mentioned, Twitter and the generated data on the social medium are not intended for scholarly purposes, it can be put to this use. For the acquisition of the data, a toolset was used that is specially created to conduct research with (Borra and Rieder 266). Twitter’s Terms of Service prohibit the use of scrapers to obtain data (twitter.com). It was therefore chosen, among other reasons, to use the DMI Twitter Capture and Analysis Toolset (DMI-TCAT) that repurposes Twitter’s Application Programming Interfaces (APIs). It is important to note that the use of Twitter’s APIs shapes the research that can be undertaken, as DMI-TCAT follows Twitter’s structure and needs to adhere to the social medium’s

regulations (Borra and Rieder 267). This means that Twitter can shape the data in ways that are unknown to the researcher (Marres and Weltevrede 10). Even though this is problematic, it should be noted that the same holds for scraping. Social media already shape the content that can be created on them and tools that allow for the acquisition of this content also

organise and shape the data (Marres and Weltevrede 11). Gitelman states that the words ‘raw’ and ‘data’ contradict each other (2). Data is always shaped by the context in which it is generated (Gitelman 5-6). It is vital to see research as a distributed accomplishment that is actively shaped by a multitude of actors (Marres 140-141).

Along similar lines, DMI-TCAT is thus also a co-shaping actor in the research process. Its usage of Twitter’s APIs implicates which data can be retrieved. For the current research specifically, its employment of the REST API is of importance. It can retrieve up to approximately 3200 of the latest tweets by a public user account (Borra and Rieder 268). In this way, it shapes the possible corpus size. Furthermore, which tweets are acquired depends on the time of retrieval. Here, it becomes apparent once more how software is an actor in the research process.

Furthermore, DMI-TCAT distinguishes itself from other Twitter data capturing toolsets by providing, apart from data acquisition, built-in options to conduct analysis as well (Borra and Rieder 269). It tries to facilitate a range of different analysis techniques. The content of the tweets as well as their metadata and other associated data is output in separate tables. This output is one way in which the toolset enables conducting a content analysis. This thesis made use of the analytical possibilities of DMI-TCAT to perform a qualitative content analysis in the case study chapter of this thesis. Once more, software plays a paramount role in shaping this research.

(27)

27

3.5 Choosing an Analysis Type

Borra and Rieder, the two developers of DMI-TCAT, identify three major methods for Twitter research. These are: ‘tweet statistics and activity metrics, network analysis, and content analysis’ (Borra and Rieder 269). DMI-TCAT was purposefully designed to enable these approaches. The toolset does enable other strings of analysis but emphasises the three mentioned earlier. This emphasis is one of the reasons for choosing from these analysis types to research Twitter in this thesis. Content analysis is the type of analysis that best suits the kind of research that is undertaken in this thesis. It is ‘a systematic, replicable technique for compressing many words of text into fewer content categories based on explicit rules of coding’ (Stemler n. pag.). Unlike the other two categories of research, it deals with the actual content of the tweets. It is, therefore, an appropriate manner to inquire into the tweets as utterances, rather than analysing how these tweets are engaged with or how different users or hashtags relate to each other to create a network.

When researching on the level of content, as was done here, two other notable qualitative analysis types could have been used. These are discourse analysis and grounded theory. It is explained here what these are and why these could have been chosen for this thesis. Afterwards, the shortcomings of these two are explained. In the end, it is clarified why a qualitative content analysis was the most suitable type to use here.

3.5.1 Discourse analysis

Discourse analysis is a close reading technique that renders meaningful the way language is used by focusing on the specific wording of the text (Marwick, Ethnographic and Qualitative

Research 118; Willig 341). It presupposes that words, and by extension images, carry

meaning that is derived from context. It is based on the idea that one’s expressions both construct and are shaped by social reality (Willig 341). As such, it lays specific emphasis on power structures, and the role language fulfils in the construction and maintenance of these (Willig 341-342). It is this weight given to power relations that would have made discourse analysis an appropriate analysis type to conduct the case study of this thesis. The way humans and non-human animals relate to each other as agents to practise a distributed form of micro-celebrity is indeed a core underlying assumption in this thesis.

However, discourse analysis does not match the scope of the analysis that was set out here. Discourse analysis is a great way to inquire into the specifics. However, it is hard to look at a more general level. As already explained, this thesis set out to explore a relatively

(28)

28 uncharted territory by combining insights from Media Studies, Human-Animal Studies, and Celebrity Studies. Discourse analysis is hard to conduct on a larger sample (Marwick,

Ethnographic and Qualitative Research 118), as it expands rather than reduces the size of the

material (Flick 11). Therefore, it did not fit the aim of this thesis, as the thesis wanted to inquire into a more general idea of micro-celebrity practised by non-human animals.

3.5.2 Grounded Theory

Grounded theory is an approach that shares many similarities with qualitative content analysis (Cho and Lee 1). Even more so, argue Cho and Lee, they seem so similar that the usage of either has been wrongly classified as the other (1). The researchers identify six similar characteristics of the two approaches: based on naturalistic inquiry; flexibility of using multiple sources of data; systematic steps in analysis; seeking themes through coding process; text to be coded into categories or themes and follow qualitative analysis

trustworthiness method (Cho and Lee 15). However, apart from these similarities, they argue, there are notable differences between the two that a researcher should be aware of. Cho and Lee distinguish six areas in which the two approaches differ, namely their philosophical basis, their characteristics, their research goals, the data analysis process, the research

outcomes and the evaluation method (15). One defining characteristic of grounded theory that sets it apart from qualitative content analysis is an iterative process of both collecting as well as analysing the data, which is called constant comparative analysis (Cho and Lee 4;

Thornberg and Charnaz 153). Rather than agreeing on a sample beforehand, as in qualitative content analysis, in grounded theory, the sample takes shape because of a subsequent need of extra material while analysing the sample that has already been created (Thornberg and Charnaz 155). Whereas this open approach to sampling does have its benefits as it takes data from multiple sites to combine into theory, it has as its weakness that it is hard to know beforehand how long the entire process will take. Since writing a thesis is time-sensitive, this is a severe shortcoming of the approach. Furthermore, acquiring multiple data sets from a variety of sources did not fit the aim of this research, which specifically focuses on Twitter; how its users can practise the concept of micro-celebrity.

In addition to this, grounded theory, because of this constant acquisition of more data, is insensitive to the way data is shaped by the actors that create and provide it. Since this sensitivity is a weighty aspect of research within Media Studies – for which this thesis was written – grounded theory was mismatched. Lastly, grounded theory has a different research goal than that was aimed for here. Its purpose, as its name suggests, is to develop a

(29)

29 substantive theory (Cho and Lee 5). The thesis aimed to categorise the data used in the case study part into different aspects of the micro-celebrity practice, which then shows how existing theory has not been extensive enough to account for non-human animals practising micro-celebrity. The aim was thus not, as it would have been for grounded theory, to create a new theory altogether.

3.5.3 Qualitative Content Analysis

Qualitative content analysis is a systematic method in which data is coded and categorised to understand its meaning (Schreier 170). Schreier outlines three principal characteristics of the method: it reduces the material, it is highly systematic, and it is flexible (Schreier 170). Qualitative content analysis, unlike discourse analysis and like grounded theory, takes a more substantial amount of material and reduces it to gain a more abstract sense of how different data entries relate to each other (Flick 11; Schreier 170). It takes a highly systematic approach that requires the examination of the acquired material to code it and create categories

(Schreier 171) subsequently.

At the same time, it is also flexible since the coding can be both data-driven, that is derived from the interpretation of the data, as well as concept-driven, that is derived from literature research (Schreier 171). This flexibility is an especially productive way of

conducting research that combines both literature and empirical research, as is the case in this thesis. In this way, it could make use of emergent coding while still being attuned to the research question and its context at hand by using a priori coding as well. In this type of analysis, the researcher creates categories during the research process of the data rather than using predefined categories. The research was still conducted with a particular framework in mind, namely that of micro-celebrity. One cannot be research without some idea as to what one is looking for. At the same time, the researcher should be open to critically assess the data that he uses (Leezenberg and De Vries 81). Qualitative content analysis embraces this in its approach. It is for all these reasons that it was opted to conduct a qualitative content analysis in the case study part of this thesis.

3.6 Conducting the Analysis

In this section, it is explained which steps were undertaken to fulfil the analysis. These have primarily followed the steps as outlined by Schreier (174), with a few minor tweaks. The overview of the steps can be found below.

(30)

30 1. Preliminary analysis of a subsample of tweets

2. Creating a codebook

3. Using the codebook on another subsample of tweets 4. Main analysis

5. Interpretation of the findings

3.6.1 Preliminary Analysis of Subsample 1

A subsample of 500 tweets of the data set of 19095 tweets in total was first scanned manually. In all instances, the tweets were opened using the common URL:

twitter.com/[USERNAME]/status/[TWEETID] to be able to see other media than just text. These are tweets from the six non-human animal Twitter accounts that were selected, posted between 26 December 2008 up until 24 April 2019. This subsample stemmed from the randomly selected 1000 tweets retrieved by DMI-TCAT, using the first 500 tweets from this selection. According to Schreier, one should aim to create a subsample that may be able to represent the whole data set (175). Therefore, a random subsample was used rather than a carefully crafted one.

Coding is the classification of text into distinct, labelled categories (Kowal and O’Connell 67). It was chosen here to create coding units based on a theme rather than a formal unit (Schreier 178). A formal unit could, in this case, have been one tweet as a defining entity to analyse and categorise. Even though this may have made analysis easier, it was not unthinkable beforehand that one tweet may be interpreted to fit in different

subcategories. To allow for maximum flexibility in this regard, the units of a tweet were separated thematically rather than treating one tweet as a distinct whole.

The strategy for the analysis was subsumption. Schreier explains that subsumption is useful for researchers that work both concept-driven and data-driven (176), which was pursued in this thesis. This strategy entails going through the material until one finds a relevant passage. It is then decided if it fits an already established (sub)category. If it does, it will be placed in that category. If it does not, a new (sub)category is made. These steps were continued until the whole sample was analysed (Schreier 176). Following these steps, a coding frame was established.

(31)

31

3.6.2 Creating A Codebook

After the preliminary analysis, I had some idea as to which types of codes and categories could be formed. Alice Marwick suggests that creating a codebook is a productive way to begin this sort of analysis. It provides some guidance to know which codes may come to the fore and how to refer to them while also being flexible enough to allow for changes of these codes while analysing the whole sample (Marwick, Ethnographic and Qualitative Research 118). The created codebook can be found in chapter 4.

The categories that inform the concept-driven part of the categories stemmed from the literature used earlier in this thesis. A list of six characteristics of the practice of

micro-celebrity was made, informed by both the introduction and literature review of this thesis. These were thus derived from the literature on the practice of micro-celebrity. Apart from the names of the labels (as already indicated below), the labels got a description that defines what the label entails and indicators that signify the presence of such a label (Schreier 176-177). To clarify the labels, a few examples of coding units are provided that are placed under the corresponding label. See chapter 4 for the analysis.

1. Intimacy 2. Ordinariness 3. Authenticity 4. Commodification 5. Fanbase 6. Popularity

In addition to these concept-driven categories, the research was flexible by using emergent coding. The latter allows creating data-driven codes by analysing and interpreting the data. In this way, a specific focus was made on the practice of micro-celebrity while being open to new ways of interpreting the data. The use of emergent coding ensured that the analysis was sensitive to the data at hand (Schreier 176). Furthermore, it was decided beforehand that an a priori category could be abandoned. This rejection was the case if there were not enough tweets that could be placed in the perspective category. In this way, the characteristics of micro-celebrity were tested on the sample to see as to what extent the current characterisation fits the sample.

Pawpular Profiles: An Exploration on the Practice of Micro-Celebrity by Non-Human Companion Animals