The linguistics of self-branding and micro-celebrity in Twitter: The role of hashtags

Twitter is a linguistic marketplace (Bourdieu, 1977) in which the processes of self-branding and micro-celebrity (Marwick, 2010) depend on visibility as a means of increasing social and economic gain. Hashtags are a potent resource within this system for promoting the visibility of a Twitter update (and, by implication, the update’s author). This study analyses the frequency, types and grammatical context of hashtags which occurred in a dataset of approximately 92,000 tweets, taken from 100 publically available Twitter accounts, comparing the discourse styles of corporations, celebrity practitioners and ‘ordinary’ Twitter members. The results suggest that practices of self-branding and micro-celebrity operate on a continuum which reflects and reinforces the social and economic hierarchies which exist in offline contexts. Despite claims that hashtags are ‘conversational’, this study suggests that participatory culture in Twitter is not evenly distributed, and that the discourse of celebrity practitioners and corporations exhibits the synthetic personalization (Fairclough, 1989) typical of mainstream media forms of broadcast talk.

culture (Jenkins, 1992(Jenkins, , 2006, whereby the communicative balance between producers and recipients was reworked, so that consumers were no longer passive but active in their co-construction of texts and dubbed 'produsers' (Bruns and Jacobs, 2006). But participation is neither neutral, nor is it distributed evenly. Instead, it is constrained by market forces and hierarchies of power that interweave offline and online contexts. Far from abandoning the neo-liberal capitalism that shaped e-commerce prior to the dot.com crisis in the early 1990s, interactions in social media contexts may enable self-promotion strategies that result in social or economic gain. The forms such self-promotion might take can vary considerably from one social media site to another. Nonetheless, visibility and attention have emerged as core properties necessary for accruing status and perceived influence.
The resulting attention economy underpins the processes of self-branding and microcelebrity (Marwick, 2010;Senft, 2008). In line with current work in discourse analysis, the processes of self-branding and micro-celebrity understand identity as discursively constructed through interactions with others (Benwell and Stokoe, 2006), where social media genres are 'technologies of subjectivity' through which the self is written into being (Ong and Collier, 2005). But self-branding and micro-celebrity place particular emphasis on the construction of identity as a product to be consumed by others, and on interaction which treats the audience as an aggregated fan base to be developed and maintained in order to achieve social or economic benefit. The promotional culture of self-branding is not an innovation of social media, but rather has emerged from the marketing literature of corporations and is also narrativized by reality television (Hearn, 2008), whereby the demotic turn (Couldry, 2002) presents fame and status as available to ordinary persons. Indeed, self-branding and micro-celebrity are forms of labour undertaken by both elite and ordinary persons in order to achieve the visibility and influence deemed necessary to achieve status or fame in the offline world. Examples include the online interactions of camgirls (Senft, 2008), use of Twitter by technology entrepreneurs (Marwick, 2010) and celebrity practitioners (Marwick and boyd, 2011), and the amateur and professional performers who now populate YouTube (Burgess and Green, 2009;Tolson, 2010).
Forms of self-branding and micro-celebrity operate on a spectrum which includes corporations who personalize their identity (Chouliaraki and Morsing, 2009), the use of branded products to signal status and identity (Marwick, 2010) and more generally, the production of a public personae that can be treated as a 'commodity sign' that is consumed and reproduced by others (Hearn, 2008). Although this self-mediation may draw explicitly on marketing techniques, as did the fictional blogger Lonelygirl15 who helped reshape the performativity of YouTube (Christian, 2009), self-branding need not be selfconsciously intentional in this way. Instead, self-branding is more generally embedded in wider trends where discourse itself has become a valuable commodity as a form of 'searchable talk' (Zappavigna, 2011), which is exploited as a resource used by marketing and advertising companies. To be made searchable is at once narcissistic (enabling the self to be found by others, and gain an audience as a sign of status) and to become the subject of surveillance (scrutinized in the service of a third party's self-promotion). The discourse contexts used for self-branding in social media genres are aptly described by Bourdieu's (1977) metaphor of the linguistic marketplace. Bourdieu's metaphor draws attention to the social and ideological nature of language use, where linguistic competence (in Bourdieu's terms (1977: 648), using language appropriately in order to command a listener) results in linguistic capital, which can, in some contexts, be converted into actual economic value. Judgements about what kinds of language use are 'appropriate' can vary considerably, and while Bourdieu stressed the value of standardized language, covert prestige can be associated with non-standard, creative or innovative forms, including those used in computer-mediated discourse (Herring and Zelenkauskaite, 2009). Given that self-branding and micro-celebrity require the individual to gain and maintain the attention of their audience, we might ask what kinds of language are deemed most appropriate for the task of promoting visibility, focusing on the resources employed within a specific virtual marketplace: Twitter.

Twitter as a linguistic marketplace
Twitter was founded in 2006, and since 2009 has been regarded as having gained mainstream Internet use. Like all websites, the composition of its membership is constantly changing, but in 2011 it was said to include 200 million registered users, with millions of updates posted each day. Similar to social network sites as defined by boyd and Ellison (2007), Twitter enables its members to construct a publically available profile, to construct and to display a list of fellow site members with whom the member shares an interest. 1 As a site for self-mediation, Twitter is ideally suited to performances of self-branding, for the default setting for profiles is public (an estimated 7% of users adopt privacy settings). Although it allows interaction between any of its members without the mediation of gatekeepers required in offline contexts, and foregrounds immediacy as a feature of its 'fresh' talk (Lister et al., 2009), Twitter is a front stage environment (Goffman, 1959) where the members' professional identity is performed (Page, 2012). Distinct from contemporary social network sites such as Linked In or Facebook, the default relationship between Twitter members is non-reciprocal: that is, if member A chooses to follow member B, this does not entail that member B will automatically gain access to member A's profile information. The asymmetric follow (O'Reilly, 2009) influences the processes of self-branding and micro-celebrity, whereby the size of a follower list is taken as a sign of status and one-to-many updates can be broadcast to an audience of potentially millions, without necessarily requiring that the updater receive updates from the audience in return.
Twitter has been described as an information-sharing site (Kwak et al., 2010), but it also enables communication between its members, and so exhibits addressivity (Honeycutt and Herring, 2009). Members communicate via short posts, known as tweets. Tweets are multifunctional, and may be used to post an update or share a link, send a public message directed to another member, or to forward a message posted by another to all the members of a follower list. Communication in Twitter is asynchronous but fast-paced, and its members have developed a number of conventions in order to keep track of the talk that emerges. These conventions include the use of the prefix '@' to signal another member's user name, the abbreviation 'RT' to indicate that a message has been forwarded (retweeted) and the use of a hashtag (#) as a prefix to indicate a search term. All three resources can be leveraged in the service of self-branding and micro-celebrity, for example, to display connection with others or to signal influence (as indicated by the number of times a user name is mentioned or a retweet is forwarded). This study focuses in particular on the potency of hashtags as a means of constructing the identity of Twitter members in such a way as to gain increased attention.
Within the linguistic marketplace of Twitter, hashtags are a crucial currency which enables visibility and projects potential interaction with other members of the site. Hashtags can be used to make a term searchable and therefore visible to others who are interested in tweets written about the same topic. When a hashtag is used with sufficient frequency, it may be listed in the 'trending topics' sidebar of the Twitter site, hence promoting a topic or term (and hence the tweets and their authors) to an audience which extends far beyond the follower list of the person who used the hashtag. For a hashtag to achieve a rank in the trending topics list is taken as a signal of status and influence. For example, fans of the pop star Justin Bieber notoriously manipulated Twitter's trending topics in order to demonstrate the popularity of the singer and the influence of his fan base (Zappavigna, 2012).
Hashtags are not only a resource for promoting visibility: they also exhibit many characteristics associated with participatory culture. Tags are created by Twitter members (rather than constructed as preselected options authorized by the site), and may be of various kinds, ranging from tags which categorize the subject matter of the tweet (in the manner of folksonomic tagging, such as #england or #worldcup) to idiosyncratic examples which function as expressive punctuation (such as internet memes like #epicfail).
Oh come on, announce the squad already #england #worldcup Female updater, Tue 1 Jun 2010 11:17 What the hell are ITV HD playing at. Complete #epicfail as England score. Missed the goal!!! Male updater, Sat 12 Jun 2010 18:37 As a search term, hashtags can be used to aggregate tweets posted by multiple members. This aggregation can create a polyphonic backchannel such as a commentary on a particular event (Reinhardt et al., 2009), or may harness the participation of a viewing audience of a television programme in real time (Anstead and O'Loughlin, 2010), such that hashtags have been dubbed 'conversational' (Huang et al., 2010). However, the conversational nature of tagging may well be less like face-to-face dialogue and closer to the para-social simulations of conversationality found in broadcast talk (Horton and Wohl, 1956;Tolson, 2010). Likewise, although hashtags are said to enable ambient affiliation (Zappavigna, 2012), that is, affiliation that is indirectly inferred rather than involving direct engagement between members, the non-reciprocal nature of Twitter networks means that this affiliation can be similarly asymmetrical and need not involve dialogic exchanges. For example, members may project their identity as affiliated within a collective group (as fans or fellow members who use the same hashtag), but they need not follow each others' accounts or communicate directly with each other.
Existing studies of hashtags have either scrutinized data samples elicited on the basis of searches for tags related to particular events, such as Barack Obama's presidential election (Zappavigna, 2011), the Winter Olympics (Gruzd et al., 2011) or large-scale corpora derived from Twitter as a whole. As yet, there has been no comparison of how particular groups within Twitter might use hashtags in similar or different ways; nor of the different types of talk in which hashtags occur and how these might be used to discursively construct identity as a social act of positioning the self in relation to others, and how that identity work is implicated in processes of self-branding and micro-celebrity. This study is a first step in this direction and compares the frequency and discursive contexts in which hashtags are used by members who might self-brand in a more or less explicit fashion: selected groups of corporate accounts, celebrity practitioners and 'ordinary' Twitter members.

Data sample and methodology
The data sample in this study consists of tweets extracted using the Twitter API from 100 publically available Twitter accounts. Forty accounts represent corporations, 30 belong to celebrity figures 2 and 30 were randomly selected as belonging to 'ordinary' Twitter members (that is, the profiles were maintained by individuals rather than a corporation, and those individuals were not constructed as famous in the mainstream media). For noncorporate accounts, equal numbers of women and men were selected (as indicated through the member's stated profile information). The tweets from the celebrity and 'ordinary' accounts were gathered in June 2010, while the tweets from the corporate accounts were gathered in June 2011. Corporate accounts represented a range of business interests:  enables the analysis to trace the discourse behaviour found in particular kinds of Twitter account, and so to explore the relative extent to which patterns of communication implied by self-branding and micro-celebrity are distributed.
The analysis that follows uses corpus-based techniques in order to establish the relative frequency with which hashtags occur in the dataset and which terms are most frequently made visible with a hashtag prefix. Manual coding was then used to identify the clause types in which the most frequent hashtags occurred, and the results of the analysis interpreted from the perspective of Critical Discourse Analysis. Specifically, I draw on the established concepts of broadcast talk associated with the Ross Priory group (Hutchby, 1996;Scannell, 1991;Tolson, 2006) and synthetic personalization (Fairclough, 1989) as models which explain trends in the discourse styles adopted in Twitter as rather less participatory and conversational than has been assumed thus far.

Frequency of hashtags and their discourse context
The data for each participant group (corporate, celebrity and 'ordinary' accounts) were categorized according to whether the tweet was an update, a public message addressed to another individual or a retweet. Analysis of the tweet types in this dataset suggest that all three participant groups favoured the one-to-many practice of updating rather than posting individually addressed public messages. In line with other studies Honeycutt and Herring, 2009), retweets remained the least frequent tweet type for all three groups (see Figure 1).
However, as the results in Figure 1 show, the relative balance between updates and addressed messages varies, suggesting different emphases on broadcast and conversational models of communication. The difference between updates and addressed messages is most pronounced for the celebrity practitioners, where updates (one-to-many posts) occurred nearly twice as often as the more conversational addressed messages (63% and 32% respectively). Conversely, this difference was the least pronounced for 'ordinary' accounts where broadcast style updates and conversational addressed messages were of broadly similar proportions (48% and 42% respectively). The relative frequency of hashtagged terms was then calculated for both the updates and addressed messages for each participant group (see Figure 2). 5 The results of this analysis show that hashtags are used to make updates visible more frequently than addressed messages. In other words, it is the one-to-many broadcasts that are the primary site of self-branding via hashtags, rather than the conversational exchanges found in addressed messages. This pattern is true for all three participant groups, but is most strongly seen in the tweets posted by corporate accounts, where the frequency of hashtags in the updates was more than 10 times greater than that in the addressed messages (12,569 hashtags per million words in updates compared with 1220 hashtags per million words in the addressed messages). Conversely, the participants most likely to use hashtags in addressed messages were the 'ordinary' updaters, who, while still using hashtags more frequently in updates, did make their conversational exchanges more visible via hashtags that did the other groups.
The hashtags used by each participant group in their updates were distinguished according to whether the tag was representing a topic (in the manner of folksonomic tagging) or whether the tag was expressing an evaluative sentiment (either an internet meme or as an idiosyncratic creation). As the results summarized in Figure 3 indicate, hashtags are primarily used to make the topic of a tweet visible, rather than to emphasize stance. Idiosyncratic or expressive uses of hashtags do occur, but these examples are by far in the minority for this dataset, and were most likely to be found in updates posted on 'ordinary' accounts, and least often in the corporate account posts. The visibility required for selfbranding appears more dependent on categorizing the updates (and hence the author) rather than on foregrounding evaluative response. We might then ask what kinds of topics occur to categorize tweets, and what identity work this achieves for the different kinds of Twitter member.

Topic-based tweets: Participatory or promotional culture?
A more fine-grained analysis of the frequency of individual hashtags is limited by the number of accounts sampled and the relatively small size of the dataset used in this study. Moreover, the frequency of a particular hashtag can fluctuate dramatically, as determined by temporal and geographical contexts for that update. For example, the hashtag #ge2010 used to refer to the general election which took place in the United Kingdom in May 2010 occurred frequently in the tweets from British updaters during the period preceding that event, but was not used by updaters from other national contexts in that period, nor is it currently used in Twitter. Nonetheless, individual hashtags can be interpreted in terms of their general semantic field (such as politics, sport and so on). Although the results that are discussed in the following section cannot be taken as representative of the entire behaviour of Twitter and are best regarded as a selective snapshot of hashtagging practice as employed by particular groups of updaters, they indicate what kinds of material were promoted as searchable for each group and used in the service of self-branding and micro-celebrity. The hashtags used in the updates of each participant group were quantified and the 12 most frequent terms are summarized with their relative frequency in Table 1. The keyness of each of the most frequent terms was verified by comparison with the Twitter reference corpus, HERMES. 6 At the surface level, there appears to be considerable diversity in the items which are marked with a hashtag in the updates. Only one tag, #FF, is shared by all three participant groups. #FF might be considered as the hashtag which illustrates most aptly the mechanics of seeking and gaining attention in Twitter. #FF is the abbreviation for 'Follow Friday', a weekly practice whereby Twitter members promote to their follower list the user names of other members that are deemed worthy of interest. These recommendations are considered a token of esteem within the linguistic economy of Twitter which enhances the visibility and potential growth of the nominated members' follower list. But while the Follow Friday practice appears in part altruistic and participatory, it also manifests subtle forms of self-branding, insofar as it enables the recommending updater to establish their position as an expert who differentiates the hierarchies of perceived value in Twitter. The list of recommended user names is one means by which the updater can display their network of contacts, and affirm their bonds within that network, which often (although not always) reflects their professional identity. For example, the department store, Selfridges, used #FF to promote fashion designers and magazines, while the actor William Shatner's 'colleagues and friends' include other actors and directors, and a practising lawyer recommended 'legal industry peeps'. #ff these legal industry peeps @username1 @username2 @username3 @username4 @username5 #law #uklaw Male updater, Fri 30 Apr 2010 13:10 Follow Friday! #ff @vogue_london @grazia_live @nicolerichie Selfridges, Fri 26 Feb 2010 14:43 Another #FF for more colleagues and friends @username1 @username2 @username3 @username4 @username5 and one more for @username6 William Shatner, Fri 23 Apr 2010 22:26 The #FF tag also appeared with expressions of thanks, which both acknowledged and reaffirmed the hashtag as a means of accruing visibility and support.
A BIG thank you to everyone who #FF, RTed & mentioned us over the weekend. We always appreciate your support! Hoover, Mon 13 Jun 2011 17:40 thanks for the #FF love @username1 @username2 @username3 @username4! Traveling, will #FF next week… Male updater, Sat 20 Mar 2010 05:10 As a form of politeness, expressing thanks implies that the nominated person named in the 'Follow Friday' tweet is in the debt of the recommender, and highlights the implied reciprocity by which recommendations circulate in Twitter. But, at the same time, posting such thanks also builds the reputation of the recommended member by projecting their identity as someone who is esteemed by others to be worth following. In some cases, the #FF practice becomes more explicitly self-promoting, where corporations and celebrities use the practice to advertise their products or outlets, such as the The other frequently occurring hashtags represent categories which allow the Twitter member to position their tweet (and so by implication their identity) within a given field of interest or expertise. Although the individual tags are diverse, they fall into two broad areas: categories related to professional identity and categories related to national events. These tags are ideological resources which position the Twitter member variously as producers or consumers in relation to the represented field, where the tags for a national event are used to aggregate the audience's public commentary and evaluation (as active consumers), while search terms related to professional expertise tend to emphasize the tweet author's identity as a practitioner within a particular field (as self-branding producers). Although there is some overlap, the distribution of the tags in these two types of category varies according to participant group. The most frequently occurring tags used by the 'ordinary' Twitter members tend to represent their contributions to commentaries which emerged around national events. These events included politics (#ge2010, #ukelection, #leadersdebate all relate to the general election in the United Kingdom in May 2010), television shows (#eurovision for the Eurovision Song Contest) or sport (#cricket, #rugby). These fields are in line with the topics which are generally discussed on Twitter as a whole (Hargittai and Litt, 2011), and reflect the growing use of Twitter as a forum for expressing the opinions of the 'viewertariat', a 'section of the audience, that aided by emerging technologies like Twitter, comments on events on the screen, responds and gives meaning to the broadcast in real time' (Anstead and O'Loughlin, 2010: 5). The prominence of these tags reflects the way in which Twitter is being put to participatory ends, making visible the reactions and interactions of the viewing audience. However, the extent to which the discourse surrounding these tags is conversational is questionable, for other studies have documented little dyadic exchange between posts aggregated as a backchannel (Reinhardt et al., 2009). Instead this publicly available response may be collated and used for political or commercial ends, such as predicting the success of a political party or a film (Asur and Huberman, 2010;Tumasjin et al., 2010). More generally, although these tags give prominence to the participatory potential of Twitter, they still position the commentators as an audience which consumes and responds to content created by others, rather than as producers who create the mainstream media spectacle which is being observed.
Hashtags for national events (such as sports competitions or television shows) were also among the most frequently occurring tags used by celebrity practitioners in this dataset. While this appears to collapse the boundaries between celebrity performance and audience response, the sports events, television shows and charity events identified by hashtags were all occasions where the celebrity practitioner was performing. Rather than positioning the authors as part of the viewertariat, these hashtags are used by the celebrity practitioners to project their identity in relation to commodities: their own performances, products and campaigns. Hence, the tags related to sports (#pdc, #grandslamdarts and #worlddarts) are from the account of the actor Stephen Fry, who also appeared as a celebrity commentator on the Premier Darts League which was broadcast through the national television channel ITV in the United Kingdom. The most frequent tags which refer to television shows in the posts from celebrity practitioners (#foodrevolution, #fallonmono) are shows in which the tweet author was the main presenter (celebrity chef Jamie Oliver, and actor and television host Jimmy Fallon, respectively). The use of hashtags by celebrities thus appears closer to the marketing strategies which characterize the updaters as producers of commodities, which they attempt to persuade the audience to interact with, be that through joining a campaign, watching a show or purchasing a product: all activities which might increase the celebrities' status in the offline world.
The products and performances promoted by celebrity practitioners in their hashtags are part of the professional, front stage identities that they perform in the context of Twitter. The importance of making one's professional identity visible as a form of selfbranding is most prominent in the choice of hashtags used by the corporate accounts in this dataset. These hashtags use company names (#EMC, #Rubbermaid), slogans (#cookwithus) and products (#Thor) along with descriptors which identify the professional field in which the corporation is an expert (#sales, #taxes). The practice of using a hashtag to indicate a field of professional expertise is by no means exclusive to corporations. 'Ordinary' Twitter members also did this, but less frequently, and never by using an employing corporation's name as a hashtag. Examples include a baker using #realbread, a priest using #synod, and lawyer using #law, #business and #corporate. The difference in the individuated company names used as hashtags by corporate accounts compared with the superordinate terms used to identify 'ordinary' Tweeters within professional communities is clearly shaped by economic power: those with greater power (corporations) can individuate their identity, whereas those with less economic capital ('ordinary' Twitter members) must affiliate themselves within a wider generic category. It is not that corporations are self-branding and that 'ordinary' professionals in Twitter are not. Clearly, 'ordinary' tweeters are also using Twitter strategically to promote their professional identities, whether that is by sharing a photograph taken at a professional event, or listing their user name within the searchable categories of the Twitter directory. But the relative frequency and degree of individuation typical of the hashtags used for this purpose by corporate and 'ordinary' groups of Twitter members suggests that this kind of behaviour is not evenly distributed and is rooted in, and reinforces, offline asymmetries of economic power and status that operate in offline contexts.

Micro-celebrity, para-social and participatory talk
The frequency and distribution of hashtags in this dataset suggests a spectrum of selfbranding behaviour, where corporations use hashtags to promote their company name and field of expertise, celebrity figures use hashtags to advertise their products and performances, while the hashtags used by 'ordinary' Twitter members construct their position as commentators on cultural events produced by others. The underlying contrast between mass broadcasting of commodities to be consumed (the company's services, or celebrity's products) and the participatory interactions of consumers (of broadcasts or events) is also constructed through the kinds of talk found in the tweets made visible by hashtags. On one hand, participatory interaction presupposes dialogic exchanges, while on the other, the processes of micro-celebrity, which treat the audience as a collective (rather than as individuals) entails a return to the dynamics of broadcast talk, where conversationality is simulated rather than dialogic. The difference between dialogic interaction between peers and broadcast styles of communication is reflected in the grammatical choices made by tweet authors, for example in the relative frequency of declarative, imperative and interrogative forms. The clause types used in the tweets for the most frequently occurring hashtags in the celebrity, corporate and 'ordinary' categories (#foodrevolution, #EMC and #leadersdebate) were analysed, and the results summarized in Table 2.
The results of this analysis indicate that for all three participant groups, declaratives were most often provided the context for the frequently occurring hashtagged terms. Given that Twitter is a linguistic market where the verbal tokens circulate in the exchange of information, this is not surprising. Corporations showed this tendency more than the other participant groups, where the self-branding of the company name occurred in announcements of breaking news, which often documented the company's success, for example as market leaders telling the 'top stories'.
#EMC Blogger Report is out! http://bit.ly/i5Mo9V Top stories today via @username1 @username2 @username3 @username4 EMC Corp, Mon 13 Jun 2011 18:03:37 #EMC cements in lead as #storage market grows http://t.co/HRg95jw /via @username1 EMC Corp, Tue 07 Jun 2011 18:31:37 Given that celebrity practitioners use hashtags to promote their products, it is perhaps no surprise that imperatives were the second most frequent clause-type for the high frequency hashtags. Typically, imperatives are used to persuade the audience to engage in some kind of activity. This construction occurred most frequently in the updates posted by celebrity practitioners and least frequently by 'ordinary' updaters. An example is the use of the tag #FoodRevolution which was used by the British celebrity chef Jamie Oliver to encourage his audience to watch his show and support the related 'Food Revolution' campaign.
The relationship between the updater and their audience as expressed in these imperatives clearly reflects the processes of micro-celebrity, where the audience is conceptualized as a fan base to be managed rather than as individual peers. The audience is referred to by collective nouns, such as 'guys', nations, 'USA', and show-specific nomenclature such as the 'Food Revolution Army', hence constructing communication from the celebrity practitioner as one-to-many interactions, rather than dyadic conversations between named individuals. While the audience is affiliated into a single group, the informality of tweet, using colloquial phrases ('HI GUYS', 'nice one'), use of expressivity, such as signing by name with kisses ('jamieox') and abbreviations typically found in 'text speak' ('2 ur battle stations') simulate the spontaneity and intimacy of 'fresh talk', especially as found in other more dyadic, computer-mediated contexts like SMS messages.
Interrogatives occurred most often in the updates posted by 'ordinary' tweeters which contained the hashtag #leadersdebate. The implied audience of the questions varies, including questions which appear to be addressed to the politicians that appeared on the programme Leadersdebate: 7 What is an illegal "imNigrant" Gordon?? #leadersdebate Male updater, Thu 22 Apr 2010 20:17 Other questions were addressed to other members of the viewing audience with whom the tweet author positions his or her self, for example by the inclusive pronoun 'we' or the general term 'anyone else'. These questions illustrate the practice of 'asking the crowd' (Zappavigna, 2012), which is common in Twitter more generally, and implies the tweet author's heteroglossic acknowledgement of a wider participating audience of other Twitter members who might affiliate with the topic highlighted by a hashtag. However, from the dataset collected, it is impossible to tell how far the questions accompanying #leadersdebate prompted responses from other members of Twitter also watching the programme. The hashtag #leadersdebate did occur in addressed messages found in the public timeline for this dataset, such as the following example. @username agree with you -I'm always amazed he doesn't use stronger language; he seems to lack conviction #leadersdebate Male updater, Thu 15 Apr 2010 20:47 But a comparison of the updates and the addressed messages shows that #leadersdebate occurred 10 times more frequently in the updates than in addressed messages (0.01 and 0.001 occurrences per tweet respectively). It would seem that the kinds of talk which aggregate around hashtags (even those used by 'ordinary' Twitter members) involve multiple participants talking simultaneously about the same topic, rather than individuals necessarily talking with each other in dyadic exchanges that resemble a conversation. This is not to say that the kinds of adjacency pairs found in face-to-face conversation do not occur in Twitter at all. Questions and answers, or statements followed by evaluative assessments did occur, even in the updates which functioned as one-to-many broadcasts. The first case, where questions followed in the same tweet by an answer, appeared as a rhetorical strategy typical of advertising used by corporations. Modified retweets are another, more innovative format in which dyadic exchanges are reconfigured as forms of mass broadcasting. Modified retweets occur when a tweet is forwarded by a second member of Twitter to their follower list, with an additional comment added. Although conventions for representing retweets are still in flux, a common practice is to add the later comment at the start of the retweet, before the abbreviation 'RT'. Thus the structure of a retweet is: [additional comment] + RT + [forwarded message]. In the following example, the original message appears immediately to the right of the abbreviation RT. The additional comment (underlined below) was added by the @ Rubbermaid account later.
GreatTips!Thanks! RT @username: @rubbermaid I recommended Rubbermaid in my TV spot about prepping for yard sales: http://ow.ly/4nM3N Rubbermaid, 28 Mar 2011 20:20 (emphasis added) Modified retweets occurred in the context of the frequently used hashtags used by celebrity practitioners, such as #Foodrevolution. Although modified retweets did not occur in conjunction with #leadersdebate and only rarely with #EMC, modified retweets were used by all three participant groups with or without other hashtags. The distribution of all modified retweets was calculated on the basis of the relative frequency of the abbreviation 'RT' in the updates.
The results summarized in Figure 4 suggest that the practice of modifying retweets is favoured most strongly by celebrity figures and occurred least in 'ordinary' accounts. One explanation for this difference is that modified retweets are used as an affiliative technique by celebrities and corporate accounts in order to project engagement with their audience as necessitated by the processes of micro-celebrity. Closer examination of the material which is added in a modified retweet by the participants in each category supports this explanation. Often, addition material from celebrity and corporate accounts is an evaluative assessment which expresses positive endorsement of the original tweet author and the sentiment they articulate.
Sweet! Keep an eye out 4 gr8 offers! RT @username: @LuxorLV next time i go to vegas i know where im staying that, it rhymes w/ huxmor LuxorLV, Mon 14 Mar 2011 05:31 Good idea! You can also recycle them! RT @username1: @username2 @Rubbermaid we put our shreddings in the compost #rubbermaid Rubbermaid, Tue 11 Jan 2011 16:45 Love this.RT @username: URDB record most fangirls to tweet jf the samething @ sametime. We l you Jimmy! Your show blows our pants off Jimmy Fallon, Sun 10 Jan 2010 05:04 The positive evaluation added by the corporation or celebrity practitioner functions as face-building relational work (Locher, 2006) which projects solidarity between the retweeter and their audience. At the same time, it acknowledges the value of the audience as satisfied consumers of the celebrity's or corporation's products (such as Jimmy Fallon's show, Dannii Minogue's dress design, Luxor's hotels) and so builds the reputation of the retweeter by reproducing the audience's endorsement of the consumed goods and their producer. The modified retweets are not dyadic exchanges (as an addressed message between two Twitter members might entail), but rather are conversational snippets that are broadcast to the millions of followers on the celebrity or corporate Twitter accounts. In this respect, they are examples of what Fairclough (1989) described as synthetic personalization, which simulates the personal address between peers, but in practice maintains the power relations between the participants. The audience's addressed messages to the corporation or celebrity may have taken place without the gatekeeping mediation of an agent or PR assistant, but their messages are used to shore up the reputation, and in turn enhance the economic status, of the celebrity or corporation. In turn, the audience and their discourse has become a commodity within the linguistic marketplace of Twitter, used to build social and economic capital of those already in a powerful position in the offline world.

Summary
The analysis of hashtags has identified the contrasting ways in which corporations, celebrity practitioners and 'ordinary' Twitter members use this as a resource to command the potential attention of an audience within the linguistic marketplace of Twitter. Hashtags are used most frequently by corporations and celebrity practitioners, who use these hashtags to make visible company names, slogans and products. These hashtags are broadcast in one-to-many updates, which emphasize declarative forms or imperatives that in turn seek to persuade the addressed audience to engage with the promoted commodity. This form of branding is clearly in line with the discourses of marketing, which use strategies of amplification to promote commodities to be consumed by others. However, forms of self-branding exist on a continuum, and are adapted with lower frequency by 'ordinary' Twitter members who similarly use hashtags to make their professional identity searchable by using hashtags which categorize their posts (and by implication, also their own identity), but promote their identity as affiliated (rather than individuated) within a wider professional field.
The promotional practices of self-branding via hashtags are offset by the interactions typical of micro-celebrity. The clause-types and choice of hashtag suggest that there is a difference between the participatory contributions of 'ordinary' Twitter members who pose questions and opinions about national events that they observe, and the para-social interactions employed by celebrity figures and corporate accounts who use the innovative formats of a modified retweet to project their identity as engaged with their audience and to endorse the value of their followers. The results imply that although the kinds of talk found on Twitter do exhibit some of the characteristics of participatory culture, it is misleading to assume that the conversational qualities of Twitter, and of hashtags in particular, mimic the dyadic exchanges typical of face-to-face interactions between peers. Instead, the talk surrounding hashtags sometimes appears closer to the qualities typical of broadcast talk, which simulates conversational qualities in the service of micro-celebrity. Twitter is a heterogeneous discourse context, in which the kinds of talk vary according to the type of Twitter member and will no doubt continue to evolve over time. Further, fine-grained accounts of computer-mediated discourse are therefore needed to explore the development of this and other virtual marketplaces as a counterpart to large-scale studies of Twitter as a whole.

Notes
1. In Twitter, those who are listed as receiving updates from another member are known as their followers. 2. In line with current approaches, I do not regard 'celebrity' as a static attribute, but rather a performed status. Although, like Marwick and boyd (2011), I agree that fame is based on a continuum, there is a clear distinction between the fame that is performed through the mechanics of mainstream media and fame achieved through online interactions alone. Hence the celebrity practitioners identified in this study are those whose activities (including their use of Twitter) were documented in mainstream media channels. 3. The tweets taken from 'ordinary' accounts were anonymized. However, I retained the usernames which occurred in examples taken from 'celebrity' and 'corporate' accounts as these referred to publicly recognized identities. 4. As counted by Antconc software (Anthony, 2011). 5. As unmodified retweets were authored in their entirety by a person other than account holders, they were not included in the rest of this study. 6. HERMES is a multi-million word corpus of tweets compiled at the University of Sydney, and forms the basis of the large-scale study of Twitter and hashtags reported in Zappavigna (2012). I am grateful to Dr Michele Zappavigna for allowing me to access this corpus. 7. The British Prime Minister at the time of the General Election in May 2010 was Gordon Brown.