Show Less
Open access

Tweets from the Campaign Trail

Researching Candidates’ Use of Twitter During the European Parliamentary Elections


Edited By Alex Frame, Arnaud Mercier, Gilles Brachotte and Caja Thimm

Hailed by many as a game-changer in political communication, Twitter has made its way into election campaigns all around the world. The European Parliamentary elections, taking place simultaneously in 28 countries, give us a unique comparative vision of the way the tool is used by candidates in different national contexts. This volume is the fruit of a research project bringing together scholars from 6 countries, specialised in communication science, media studies, linguistics and computer science. It seeks to characterise the way Twitter was used during the 2014 European election campaign, providing insights into communication styles and strategies observed in different languages and outlining methodological solutions for collecting and analysing political tweets in an electoral context.

Show Summary Details
Open access

3. Exploring the dialogical dimension of political tweets: A qualitative analysis of ‘Twitter styles’ of UK candidates during the 2014 EU Parliamentary Elections (Kondrashova, Tatiana / Frame, Alex)

Tatiana Kondrashova & Alex Frame,

Univ. Bourgogne Franche-Comté

3. Exploring the dialogical dimension of political tweets: A qualitative analysis of ‘Twitter styles’ of UK candidates during the 2014 EU Parliamentary Elections


This chapter looks more closely at « informational » and « interactive » communication styles on Twitter, qualitatively examining the multiple and contrasting ways in which different functional operators can be used, and proposing a model based on the “dialogical potential” of the tweet, which aims to avoid some of the more reductive generalisations commonly found.

3.1 Introduction

Since Twitter has become widely used by politicians in many countries, much has been written about the different ways in which this tool is, can and should be exploited (Frame and Brachotte 2015; Jackson and Lilleker 2011; Larsson and Moe 2013; Parmelee and Bichard 2013). Many researchers have denounced a general tendency for politicians to try to reproduce one-to-many communication strategies, tweeting at their voters rather than seeking to ‘interact’ with them in what is often presented as a more appropriate ‘web 2.0’ style (Small 2010; Vergeer and Hermans 2013). Further attempts have been made to classify communication styles on Twitter, generally ranging between the two poles of one-to-one (interactive) and one-to-many (informational) (Dang-Anh, Einspänner, and Thimm 2012; Eyries and Poirier 2013; Sæbø 2011). This chapter will build on existing approaches to study to what degree the use of different operators on Twitter appears to be correlated with different “Twitter Styles” (Dang-Anh, Einspänner, and Thimm 2012). Adopting a qualitative, context-based approach to analyse messages sent by UK candidates at the European Parliamentary Elections in May 2014, we will seek to further characterise the uses and perceived communicative functions of the Twitter-specific communication operators (@, RT, #, http) which contribute to making the micro-blogging network a specific discursive space (Dang-Anh, Einspänner, and Thimm 2012). Anne-Marie Paveau argues that Twitter discourse, as a specific ‘techno-discursive’ production can be seen to←53 | 54→ constitute a new genre: “Twitter writing” (Paveau 2012), while Michele Zappavigna sees the discursive dimension of the platform as the key to understanding its “networks of affiliation” (Zappavigna 2012).

Interactions between users on Twitter are the bricks and mortar of the web-based community, the Twittersphere, which, from a technical if not a social point of view, has high potential for generating exchanges in political communication (Ausserhofer and Maireder 2013; Thimm, Einspänner, and Dang-Anh 2012a). When studying online interactions or computer-mediated communication (CMC) it is important to take into consideration interpersonal social interactions (Goffman 1967) but also the specifics of the medium, the technical devices used and their functions (Coutant and Stenger 2010). Compared to face-to-face interactions, the absence of non-verbal means of communication like gestures, mimics, intonation etc. should be taken into account, as well as potential erasing of social distance between communicants, time and place indeterminacy (Bucher 2004). At the same time, the construction of a dialogical situation could be compared to offline communication in terms of the background social and cultural knowledge required to produce and interpret symbolic acts. Language skills and knowledge about the structure and functions of the mediating technical devices are necessary to “release the interactive potential” of the interface (ibid.).

In our approach, we do not limit interactions to face to face exchanges (Goffman 1967), but extend this notion to all human activities in which two or more actors are involved (Vion 1992), at least potentially. Publishing tweets in the Twittersphere involves one17 or more actors, and may create a constellation of “interactants” (Charadeau and Mainqueneau 2002), if we take into account the potential audience of the message posted, or indeed the ‘conversations’ it generates. The ‘presence’ of face-to-face interactions is recreated in CMC by various means by which the device allows users to organise their discourse (Bays 1998). Non-verbal interactions are compensated for by other elements inherent to the device’s technical structure: in the case of Twitter, these are the communication operators. As Caja Thimm and colleagues have pointed out, each sign becomes a communication operator by taking on different functions within the context of the Twittersphere, as characterized here:←54 | 55→

Figure 3.1 Classification of communication operators (Thimm, Einspänner, and Dang-Anh 2013)


Based on their analysis, Thimm and colleagues distinguish two major “Twitter styles”: the interactive-personal style, characterised by tweets which invite interactions and interpersonal exchanges, and the thematic-informative style, more concerned with distributing information (Thimm, Einspänner, and Dang-Anh 2012b). The former is characterized by an abundant use of @ and RT operators, while the latter favours external links. In order to simultaneously question and further characterise these styles, this chapter focuses not only on the communication operators, but on their functions as deployed in a specific tweet which can be interpreted in its technical and communication environment and in relation to a wider social and political context. We can ask, for example, whether a journalist who systematically adds to her professional tweets the Twitter handle of the newspaper for which she works, @Telegraph, should accurately be described “interactive”: she is most probably not seeking a reaction from the newspaper but looking to give her tweets more visibility while underlining her professional affiliation. Similarly, if someone tweets a link in order to provoke reactions from friends, are they employing a solely “informational” strategy? This chapter will endeavour to propose a heuristic framework allowing us to better take into account such specificities.

3.2 Twitter styles revisited

In order to further characterize the Twitter styles, by building on various approaches and trying to limit the overall number, we can list a certain number of←55 | 56→ partially overlapping major communication functions: addressing, mentions or referencing, contextualisation, argumentation, and passing on information. These can then be related to the communication operators, as presented in Figure 2: each communication operator posesses a main function but could partially fulfill the adjacent functions as well.

Figure 3.2 Five major communication functions correlated with operators


A slightly different way of thinking about the interaction/information continuum is to consider the potential reach of a tweet. A tweet’s reach is defined as the number of accounts to which it is potentially visible. A tweet sent by an account (x) becomes potentially visible to another (y) when any of the following conditions are fulfilled:

(a) the account owner of y follows account x;

(b) the tweet sent by x contains the name of account y (an “@ mention”);

(c) an account followed by y retweets the tweet sent by x;

(d) the tweet sent by x contains a # which y searches for (or follows using a specific tool);

Some tweets, for example addressed by one account with relatively few followers to another, may have very low reach, potentially being seen by only a few other users. On the other hand, a tweet including very popular twitter handles, plus common or trending hashtags can potentially be seen by millions. Although a lot depends on the actual contents and context of the tweet itself, the functions←56 | 57→ previously listed can be roughly correlated with their potential reach or the visibility they may give to a particular tweet, as resumed in Figure 3:

Figure 3.3 Relative potential reach of major communication functions


In other words, depending on a communication operator and its contextual function, as well as the element with which the operator is directly combined, its neighbouring elements and their occurrences and significations in a wider context, we can consider that a tweet will have greater or lesser potential reach. It is the technical nature of the operators and the way they function in this particular medium, in relation with the communication environment, that allows us to talk of a ‘techno-discursive platform’, where the technical elements mobilised in a given interaction contribute to produce pre-programmed dialogical configurations. Their previously-accumulated knowledge about how these technical elements function will allow Twitter users to employ them to seek to achieve certain functions and effects within the Twittersphere. This is close to the dialogical situations encountered in offline communication where participants seek to anticipate potential acts and utterances from their partners, and indeed this conscious or unconscious intersubjective anticipation takes place alongside more technical calculations within the Twittersphere too. Whether the partner is fictive or virtual, the way discourse is constructed is guided by principles of dialogue. From a Bakhtinian perspective (Бахтин 1996), dialogism is a characteristic phenomenon of all speech, even in the case of a monologue.

Indeed, political tweets correspond completely to Bakhtin’s logic: the meaning attributed to any tweet is linked to the accumulated tweets of the Twittersphere, and politicians try to take into consideration their potential audience(s). By looking at the way tweets are constructed, we can reason in terms of Bakhtin’s logic of the double dialogical extension of discourse (Бахтин 1996). This concept has been←57 | 58→ used to analyse political discourse online (Bonhomme and Stalder 2006), and allows us to bring together all of the elements that favour a (supposed) dialogue online: producer, potential receiver, different codes of communication (e.g. @, # in case of Twitter), other discourses. Figure 3 (supra) can be used to illustrate the different degrees of dialogism which can potentially be associated with different communication functions, and to underline the applicability of this concept to the Twittersphere. In our analysis we will use it in order to evoke the degree of involvement of a Twitter user in a potential dialogue through a tweet. Following on from this discussion, we can formulate two hypotheses:

1. “Twitter styles” are not determined only by the prevailing number of communication operators in a tweet but especially by the correlations of operators’ functions as actualized in discourse (the tweet and its wider context).

2. Based on the concept of the dialogism applied to Twitter (Figure 3), we hypothesise that the more determined a potential receiver is (low reach), the higher the degree of implication in a dialogue (dominantly interactive tweet) and, conversely, that the larger the potential audience (high reach), the less the dialogical function is actualized in a tweet (dominantly informational tweet).

These hypotheses were tested by applying them to the corpus of tweets sent by UK candidates during the European Parliamentary Elections, collected within the framework of the TEE2014 project.

3.3 Methodology & results

The corpus consists of all of the tweets of the 309 Twitter accounts which were identified as belonging to UK candidates for the European Parliamentary elections on 22nd May 2014, sent in the three-week run-up to the Elections and in the week following the vote (1st May to 27th May). 72 860 tweets were collected overall. From this corpus, we selected all candidates who had produced more than 50 tweets during the period, this being considered a minimum number for us to be able to usefully characterise the Twitter styles of a particular candidate. This query resulted in a list of 213 candidates. The final step was to reduce this list to a smaller number of candidates enabling us to carry out qualitative analysis to gain insights into the way particular operators and functions were being used. In order to obtain a varied selection of accounts, we chose to take the two candidates whose tweets had the highest percentage of each communication operator, the two candidates with the lowest percentage and two balanced accounts which←58 | 59→ seem to have close to average numbers of each communication operator (n=18 accounts).18 The profiles of these accounts are as follows

1. Abundance of communication operators: highest percentage19 of media20, RT, http, hashtags and @ (8 accounts).

2. Balanced accounts where the number of each communication operator varies between 25 and 60 percent (2 accounts).

3. Lowest percentage of communication operators (8 accounts).

The first fifty tweets collected for each account were analysed (n=900 tweets). The accounts selected are presented in Table 1, along with the number of communication operators used by each candidate, and the percentage of tweets in which each communication operator was used:

Table 3.1: General information about candidates and statistics about communication operators used


The candidates are classified by number of total tweets produced during the period of analysis. The selection criteria resulted in a fairly even spread of parties←59 | 60→ and constituencies being retained, although candidates close to the top of their particular list appear to be overrepresented. It should be highlighted that in some cases (@LindaMcAvanMEP) the abundance of one communication operator is associated with high use of another (high rate of RT and @, > 60 %), while in other cases (@KSillsEngDem) there is a considerable preponderance of one type of communication operator (almost) exclusively. Several accounts fulfilled several criteria at the same time, for example that of @EnzaFerreri had the highest percentage of http, as well as the lowest of # and @.

Discourse analysis of the 900 selected tweets was carried out in the following steps:

1. A communication operator (@, #, httpor RT) was selected.

2. On the basis of the influence of neighbouring elements on the communication operators’ function, operators were manually coded using Nvivo, depending on their accompanying element; For example, @Telegraph = @+name_of_media_source.

3. The whole syntagm was analysed, including the neighbouring elements such as prepositions, names of other people, adverbs or other communication operators that could define, change or complete the first function.

Once coded, the tweets, the communication operators and their functions were analysed manually, while Nvivo was used to give insights into word frequency.

3.4 Results and Analysis

3.4.1 The operator @

We began by seeking to establish the degree to which the operator @ can be seen as marker of an “interactive style”. A first set of tweets were coded as @+name of a personal account. It should be noted that some accounts, such as @Skooshbag, are less transparently personal accounts than others which simply prefix the @ operator to the first name and surname. Most of the time, the account profile description allowed us to clarify this. These tweets often seem to show @ in its primordial function of addressing:

(1) @meljomur @votingyes @Skooshbag @Celebs4indy I wasn’t there unfortunately but this post covers general feedback Yzihz21←60 | 61→

(2) @votingyes @Skooshbag @meljomur @DRossborough @Celebs4indy all the more reason to take the positive case to these areas.

(3) @roseleaarran me too!

The first tweet has a neighbouring element http(1) to introduce a supportive argument, the second (2) and third (3) ones obviously have a responsive character. The second tweet illustrates various types of accounts associated with an @ operator, with different degrees of dialogical potential depending on the nature and character of the account: @votingyes = @+name of community, @Skooshbag= @+name of a personal account. From a political communication perspective, this type of @ also includes politicians’ personal accounts: @+name of a politician, as in (4) and (5).

(4) RT @CllrAlexander: Great turnout for Dundee…

(5) @CllrAlexander this for talks as a rally or a big work session?

In the fourth case @ is arguably not a simple marker of interactivity any more (4), since its function has been changed through the presence of another communication operator RT. The goal of this retweet would appear to be referential rather than to enter into a dialogue, though in practice, since the original account is automatically notified of any retweets, this can also be considered as a particular form of dialogical interaction. The counter-example of a tweet using the same operator @+name of a politician (5) but accompanied by a question and not a RT radically changes its character and it can much more easily be seen as a part of a dialogue.

It is interesting to note that there are several cases of non-use of the operator @+name of a personal account, where opponents refer to the name of the opposite party’s member without using the corresponding Twitter handle (6). In contrast to this tendency, political allies are more likely to be introduced directly with @, for instance (7), addressing or referencing a fellow member of Scottish National Party.

(6) So Farage thinks all Romanians are people smugglers & Tory MEP cand Ian Duncan thinks Scots Chinese all eat cat. Fundamentally flawed people

(7) @ToniGiugliano on why we must be independent in Europe: #ep2014

It is unclear whether the use or non-use of a communication operator here is premeditated or just fortuitous, as counter-examples exist, yet it would appear to make sense as a political strategy designed both to expose one’s own message to a sympathetic public which does not necessarily follow one’s own account, or indeed to avoid a hostile audience in the case of non-use of an opponent’s Twitter←61 | 62→ handle. It could also be seen as a way to draw attention to an ally’s account, but not to that of a political opponent.

The operator @ was very frequently used in its so-called “gratification function” when accompanied by a neighbouring element such as “thanks” (8) (9) or in its localising position accompanied by a preposition of localisation “with” (10).

(8) @AmranHussain @LondonLabour @ClaudeMoraesMEP @Lucy4MEP @SebDance Very many thanks Amran!

(9) @DaveCharnley1 @AnneFairweather @ClaudeMoraesMEP @Lucy4MEP v many thanks Dave

(10) Campaigning with @BarnetLabour & going to Hale ward with @sebdance @giampialhadeff

The construction @+name of a personal account + thanks, with, etc. could not be considered as completely non-interactive, because the person in question is addressed (8) (9) or at least mentioned (10), nevertheless its dialogical potential would appear to be relatively lower than in cases of questions or answers (3).

It should be highlighted that constructions of the type @+name of a tweeting candidate retweeted by the candidate him/herself are widespread in the political Twittersphere.

(11) RT @redstarneil: Thank u @maryhoneyball et al 4 supporting @QPLabour today. Look forward 2 working with more of you! #labourdoorstep http:/…

The degree of the implication in a dialogue appears here to be low (11). The function of this communication operator, in relation with its neighbouring elements, seems to aim more at gaining visibility rather than interaction. In a similar function we find another type of @: @+party name.

(12) Did Labour and the Conservatives do well for the North East 2006–2012? Don’t vote for more of the same, vote @UKIP.

In the example (12) the dialogical potential appears seriously limited, and the tweet seems to aim rather at increasing the visibility of the party in question by mentioning it, or indeed at attracting new followers to the author’s own account by showing party affiliation. This is an example of a more general strategy which can often be observed.

The last category of @ that we would like to emphasise in this study is @+name of media source that seems to reveal the polysemic potential of this communication operator’s functions.

(14) Great start for Louise Bours on Question @bbcquestiontime!

(15) @itvnews There’s a revolution happening. Many Londoners are planning to #VoteAnimalWelfareParty this Thursday EU elections. Are u following?←62 | 63→

(16) “@TheEconomist: Losing America cost King George III his sanity. Mr Cameron thinks losing Scotland would not even cost him his job” losing??

This type of @ usage often appears more informative than interactive. It can be used for argumentation purposes (15) (16) or for passing on information (14).

Through examples 1–16, the @ operator can be seen to potentially fulfil various functions, with different degrees of implication in dialogue. This depends notably on the neighbouring elements that introduce this operator, and the nature of the account with which it is associated: name of a personal account, name of political party or organisation, or name of media.

3.4.2 The operator RT

This communication operator is by nature heteroglossic in Bakhtin’s sense, it refers to the words of others, and can be used to pass on information, sometimes accompanied by argumentation. Once again, its exact nature and function are strictly dependant on its context of usage.

In the cases where RT introduces a tweet from a personal account (1), it fulfils its referencing function as a mention that possesses some dialogical potential, and may imply a community of interactants.

(1) RT @tobyperkinsmp: Great to join @WMLabour Euro candidate @ansar_ali_khan & Sir Albert Bore to see Labour investment on Metro extension ht…

Retweeted tweets may also themselves be retweeted (2), possibly with additional comments or reactions, creating a chain of interactions and presenting an interesting case of an enlarged discursive sphere.

(2) RT @ChorltonGreen: RT @ranty_man: Hunt is desperately trying not to alienate former Labour voters #BBCQT; is it working?

Other examples show that when RT introduces the parties’ speech (3) (4) (5), it appears to lose much of the dialogical potential of the examples above (1) (2). The goal of such retweets is generally to pass on information, assure parties’ visibility or associate oneself with the particular point of view, event, etc. contained in the original tweet. This kind of tweet seems to be less engaging, given that party tweets themselves are generally more neutral that personal ones.

(3) RT @LabourStafford: Euro campaign launch w/ @neenaformep @sionsimon @LW4WM @ansar_ali_khan @katevoteslabour @annieinclover @rowan_draper: h…

(4) RT @AnimalsCount: Results for London 21,092 for #AnimalWelfareParty. Sadly, no seat for us this time. Huge thanks to our amazing supporters…

(5) RT @AnimalsCount: @Djurensparti Best of luck for #EP2014 elections tomorrow. Let’s Make History For Animals! #EU2014←63 | 64→

The last case that we will distinguish in this study is the retweet of a media source (6):

(6) RT @Daily_Record: SNP EU candidate @TasminaSheikh outlines why she and many other Scots-Asians are voting Yes in the #indyref…

This type of tweet can fulfil both functions of argumentation and contextualisation, but often in the form of quotation.

There are numerous cases of retweets of messages sent to candidates (analysed in the previous section), as shown in example (7), in order to give them more visibility or to show the candidate being addressed in a positive light.

(7) RT @ARUKnews: @maryhoneyball Looking forward to the results of the EU elections. Thanks for your support for medical research #EU4Research

A quote thus deprived of a dialogical dimension can be seen as an instrument of political approbation and argumentation and indeed this dimension is also present in the cases of political parties (3) (4) (5).

3.4.3 The operator #

The hashtag is a specific sign that constitutes an important contextualizing element of Twitter discourse, characterized by its terse style. There are several types of hashtags that can be distinguished according to content: #+phrase (1) (a syntactically-structured utterance) or #+word-phenomena (2) (including acronyms and abbreviations).

(1) Is London ready to #VoteAnimalWelfareParty on Thursday at #EP2014 elections? Pls RT if you think yes!

(2) RT @GlenisWillmott: Labour MEPs fight for LGBT rights in the UK, EU and worldwide #EP2014 @LGBTLabour @LGBTintergroup

However, as in the case of the @ operator, it is not the type of hashtag that determines its function, but the elements it introduces. Both hashtags in the first example (1) present reduced verbal units that aim to contextualize the discourse. The first conveys ideas linked to a particular cause, and the request to retweet aims to encourage people to show their support for this cause. The second might serve to increase visibility of the tweet to all people following the official #EP2014 hashtag. The potential reach of these hashtags is relatively high both because of the wide audience reached by the second hashtag and by the incitation to fellow supporters to retweet the message. The same hashtag used in the second example (2), in a retweet, seems to be more about passing on information, as well as associated references or argumentation through the link.←64 | 65→

The dialogical potential of a hashtag also varies depending on its environment. Here are a series of examples using #VoteAnimalWelfareParty. The degree of implication in a dialogue is linked to the construction of the tweet:

(3) @leonalewismusic We want to write a new chapter of history for animals at EU elections this Thurs. Will u join us? #VoteAnimalWelfareParty

(4) @itvnews There’s a revolution happening. Many Londoners are planning to #VoteAnimalWelfareParty this Thursday EU elections. Are u following?

(5) This Thursday Londoners have the opportunity to write a new chapter in this history of mankind. Play your part. #VoteAnimalWelfareParty

(6) We need around 140,000 votes from across London to make history for animals on Thursday! #VoteAnimalWelfareParty

(7) Believe in a better world for all animals? RT this: #VoteAnimalWelfareParty EU Elections May 22

In these cases, the author tries to appeal to the audience by accompanying the hashtag with linguistic elements such as questions (3) (4) (7) and imperative forms (5). The direct address is a way of trying to increase levels of engagement.

(8) #bbcnews BBC commentators and reporters cannot seem to hide their obvious enthusiasm for UKIP

(9) #bbcnews why is it the bbc are avoiding talking about German results? Maybe it doesn’t fit today’s agenda as Euro skeptics only got 7 %

The hashtags that name the information source (8) (9) can be compared with previous examples using the operator @+name of a media source. The difference would appear to be that using @ is a way of seeking to interact with a media source, whereas the choice of # sets up a conversation about the media source, often to criticise it, which can possibly be seen as an invitation to dialogue, either with the representatives of the media source, or among Twitter users who may or may not agree.

The dialogical potential of a hashtag thus depends very much on the hashtag itself: a common hashtag will add visibility to a tweet, but possibly not create much interaction, whereas a more obscure one can be a way of targeting a message to a small group of insiders who don’t necessarily follow one’s account.

3.4.4  Use of media

When considering use of media in tweets, we can distinguish videos, photos and screenshots.22 In this chapter we focus solely on the use of photos which are←65 | 66→ the most common type of media in our global corpus. Photos allow politicians to illustrate their messages and to reinforce their impact on the audience, and can be perceived as having high argumentative and interactive potential, even if this has not been established empirically in the framework of the current study. We can distinguish apparently “spontaneous” photos and others which are more evidently “staged”. Examples of photos within the first category concern some sports events, photos of natural scenes or of political debates. The aim of this type of photo seems to be to create a degree of complicity with Twitter users who could potentially be witness to these events. Staged photos often concern events from the campaign trail, possibly aiming to allow absent followers to feel present alongside their candidate.23

In individual tweets, the most common way for politicians to introduce photos is: introductory utterance + spontaneous/staged photo. The dialogical potential of a photo seems to depend on the nature of this photo and on the content of the introductory utterance.

Photo 1


This post (Photo 1) presents a composition of an introductory utterance that contains two @ communication operators: with + @ + name of politician and @ + company name and a staged photo, possibly in the hope of being retweeted, since the dialogical potential seems otherwise to be relatively low.←66 | 67→

Photo 2


The second tweet also contains an @ communication operator: @ + football club, but takes the form of a spontaneous photo this time. By choosing this particular @ mention and showing followers of the football team that he was present at the match, it can be assumed that this candidate was seeking to underline their shared allegiances, foregrounding his role of “ordinary” football supporter rather than his role of candidate.

It is interesting to note that the photos posted tend to constitute an integral part of the utterance (nominative sentence completed by Photo 1, elliptic sentence explained by Photo 2). Once again, the complementarity between the introductory utterance and the posted photo underlines the necessity for these two elements to be analysed together. Nominative sentences as illustrated by Photo 1 are recurrent:

(3) Great session with the best team in Birmingham in my own ward WWH + staged photo

(4) Lovely afternoon in Stafford with super PPC @kategodfrey @sionsimon @neenaformep @LW4WM and Stafford councillors + staged photo

(5) Fantastic 1st day of @balmoralshow – great conversations with people from right across NI & beyond. Roll on tomorrow + spontaneous photo with a political actor.

Another common type of introductory utterance: gratification element (thank you) + spontaneous/staged photo, # + spontaneous/staged photo, as for example

(6) Best wishes to the hundreds of cyclists that just passed my office in Lisburn who are #cyclingagainstsuicide + spontaneous photo of a sports event.←67 | 68→

The spontaneous photos may seem to be more interactive because they can more easily be shared and understood (2) (5) (6), while the staged photos (1) (3) (4) seem to target Twitter users curious about a precise event. In all cases, both spontaneous and staged photos can be used to convey information or arguments. Their interactivity can be measured only in conjunction with the analysis of an introductory utterance that determines the targeted audience.

3.4.5 The use of links (http)

A hyperlink is a flexible communication operator whose function and associated dialogical potential depend very much on its neighbouring elements, as the content of a shortened link is generally not immediately apparent. The vast majority of links are coded by clickable shortened URLs, which provide few clues to their target or content. This may or may not make people curious to click on them. Sometimes the content of the link is directly integrated in a tweet (for example videos from Youtube (1, below)). Politicians’ links often lead to personal or party sites, blogs or Facebook pages, whose objective appears to be to invite people to adhere to their ideas.

There are different ways of introducing (a) personal links that lead either to personal sites (3), Facebook pages or profiles, (4) (5) (6), or blogs (2), which can be termed or (b) external links leading to external sources (1) (7) (8). The dialogical potential of these links depends in the first place on the neighbouring elements: other communication operators and/or other integrated elements (videos/clickable screenshots of media web pages). Tweets (1) and (2) below illustrate this. Tweet (1) includes an external link to YouTube, yet the whole tweet seems to be more interactive than the second one (2) that leads to a personal link. In spite of the presence of a gratification element, tweet (2) does not seem to address followers in particular, and is incomplete, with followers being implicitly encouraged to click on the link for more information. The first tweet (1) contains a hashtag and a question that seem to increase its dialogical potential, in association with the integrated video.

(1) Believe in a better world for all animals? RT this: #VoteAnimalWelfareParty EU Elections May 22 + integrated video

(2) Thanks: A big thank you to everyone who took the time and trouble to vote BNP yesterday. I would also like to…

(3) The second in my week-long series of reasons to vote UKIP on Thursday. Send them a message!…←68 | 69→

Examples (5) and (6) also contain personal links that lead to Facebook but seem to reveal higher dialogical potential than tweet (1) thanks to the neighbouring elements used, such as @ + personal account name (5) or question + integrated article (6). Even an external link to a media source as in examples (7) and (8) could be seen to be interactive depending on the neighbouring elements as @ + personal account name + integrated picture (8) or opinion + external link + integrated video (7).

(4) This is the last day before the vote to elect Members of the European Parliaments for Britain, and I cannot…

(5) @meljomur @votingyes @Skooshbag @Celebs4indy I wasn’t there unfortunately but this post covers general feedback…

(6) Had enough yet? The system is broken. + integrated article from Guardian

(7) Didn’t see it on tv news at all but at least @BBCScotlandNews have the pensions confirmed safe article posted on line…+integrated video

(8) @ToniGiugliano on why we must be independent in Europe:… #ep2014 + picture

In general we can note several strategies used by politicians in order to incite Twitter users to follow their personal sites or external links: imperative phrases (3), broken sentences (4), questions (6) or even direct addressing (8). The dialogical potential of a link does not depend on its nature but on the introductory utterance and the elements integrated afterwards. A link to Facebook or another social media account may also lead sympathetic Twitter followers to follow that account too, thus increasing the candidate’s social media visibility and reach in general. The dialogical potential of tweets including social media links thus exceeds Twitter itself, since comments may be left on other platforms, after clicking on a social media link in a tweet.

3.5 Discussion and conclusions

The examples cited of the different communication operators and their functions tend to confirm the hypotheses formulated above. Hypothesis 1, concerning the importance of neighbouring elements in determining a communication operator’s function can be validated thanks to the data studied. This suggests that Twitter style classifications should be based not only on the number of each type of communication operator used by a politician, but on the functions of the prevailing operator(s) as actualized in discourse by means of its accompanying elements. This in turn entails close, qualitative analysis of the tweets in question.←69 | 70→

The second hypothesis, concerning the degree of implication in dialogue, is also supported by the data considered in this study, although the notion of “dialogue” clearly merits further investigation, which could be carried out by examining the retweets and conversations provoked by a given tweet. Despite this, the notion of implication or reach appears to be a meaningful criterion on which to differentiate tweets qualitatively, ranging from those whose dialogical markers (addressing, questions, exclamations…) and communication operators suggest that they are aimed at one or a few individuals, to those who seem to target a much wider audience. These differences in potential reach often seem to correspond to different strategies, ranging from “authentic” one-to-one dialogue – that Holy Grail of political communication on Twitter when one reads certain analyses – to the use of hashtags or at mentions which seem aimed to ensure a tweet’s visibility rather than to provoke direct exchanges with other users. There again, the contents of the tweet (more or less controversial, surprising, inspiring, etc.) and its dialogical markers, as well as the social and / or political context in which it is sent, all need to be taken into account in order to further characterise its dialogical potential.24 In order to analyse this potential, we propose the following table, applied to different operators, to resume, qualify and exemplify the differing degrees of dialogical implication.

Figure 4: the “I to I” Model2526


←70 | 71→


The acronym in “I to I Model” stands for “Information to Interaction”: often presented as the two extremes of a tweet’s potential functions. The model characterises the relationship between four key elements. (i) The communication operator’s function as mobilized in the tweet which defines (ii) the audience potentially concerned by that tweet. According to the audience potentially addressed or actually involved in a conversation, we can talk about (iii) the reach of a tweet. Based on the reach, (iv) the relative dialogical potential of the message can be estimated. This estimate is not numerical, but simply relative to other contextualised configurations of operators. The figure suggests that the degree of implication in a dialogue is inversely proportional to the number of Twitter users concerned by a tweet. Once actualized in a tweet, communication operators should give a key to understanding a politician’s strategy to reach his or her targeted audience.

Although the “I to I model” attempts to associate particular occurrences of operators and their neighbouring elements with different degrees of implication←71 | 72→ in dialogue, we have already noted that this may vary considerably depending on other elements in the tweet and on the wider context. The model thus needs to be further developed and adapted to take into account these factors. Other limits to this preliminary study include sample size and representativeness, and the particular domain (political communication) from which the tweets were chosen. While we would argue that the general principle of taking into account the context of operator use remains equally valid for tweets from other domains of social activity (corporate communications, celebrity fandom…), it is clear that each of these domains is likely to reveal its own specificities.

Future developments of this model will go further into taking into account the use of embedded or linked media (photos, videos) in tweets, as this is a widely-used feature of the platform. The approach also needs clarify how images relate to questions of dialogue and reach.

For all of its limits and the exploratory nature of this approach, we suggest that the “I to I” model opens up new possibilities to let us systematically investigate and characterise the dialogical qualities not only of single tweets, but also, by extension, of individual accounts, political parties, etc., in a more sophisticated way than many of the approaches currently presented in the academic literature. In this respect, we hope that it may help shed light on the ongoing debate in political communication into the relative “(im)maturity” of politicians’ use of the medium and the frequent denunciations of their “web 1.0 mentality”, whereby they prefer to broadcast information rather than interact. Looking at the relative reach or dialogical potential of their messages may enable us to distinguish more subtly between different strategies, taking into account the different communication functions which specific tweets might seem to fulfil, and the individuals or groups which they ostensibly target, be they political colleagues, journalists or media, activists, citizens, etc. If direct two-way politician-voter interactions may indeed be scarce, with whom do politicians more readily interact and what seems to motivate the different types of tweets they use? The answers to such questions will hopefully give us a more accurate vision of various logics of usage by politicians, which may then be fed back into the wider discussion about the possible construction of a deliberative digital public sphere and the degree to which the adoption of Twitter and other social networking sites by politicians may or may not renew democratic processes.←72 | 73→


Ausserhofer, Julian / Maireder, Axel: ‘National Politics on Twitter: Structures and Topics of a Networked Public Sphere’, Information, Communication & Society, 3 (16), 2013, pp. 291–314.

Bays, Hillary. ‘Framing and Face in Internet Exchanges: A Socio-Cognitive Approach’. 1998,, accessed on 09.09.2015.

Bonhomme, Marc / Stalder, Pia: ‘Analyse Sémiotique Des Pages D’accueil Des Sites Politiques Suisse’, La politique mise au net, no. 80, 2006, pp. 11–23.

Bucher, Hans-Jürgen 2004: ‘Online-Interaktivität –ein hybrider Begriff für eine hybride Kommunikationsform. Begriffliche Klärungen und empirischen Rezeptionsbefunde’. In: Bieber, Christoph / Leggewie, Claus: Interaktivität. Ein transdisziplinärer Schlüsselbegriff, Frankfurt : Campus. 2004, pp. 132–167.

Charadeau, Patrick / Mainqueneau Dominique: Dictionnaire D’analyse Du Discours. Paris: Edition du Seuil. 2002.

Coutant, Alexandre / Stenger, Thomas: ‘Processus Identitaire et Ordre de L’interaction Sur Les Réseaux Socionumériques’, Les enjeux de l’information et de la communication, 1, 2010, pp. 45–64.

Dang-Anh, Mark / Einspänner, Jessica / Thimm, Caja: ‘Mediatisierung Und Medialität in Social Media: Das Diskurssystem „Twitter “ ’. Sprache Und Kommunikation Im Technischen Zeitalter: Wieviel Internet (v) Erträgt Unsere Gesellschaft? 2, 2012, pp. 68–91.

Eyries, Alexandre / Poirier, Cassandra: Alexandre Eyries et Cassandra Poirier, « Une communication politique 2.0 », Communication [Online], Vol. 32/2 | 2013, accessed on 14.09.2015.

Frame, Alex / Brachotte Gilles: ‘Le Tweet Stratégique: Use of Twitter as a PR Tool by French Politicians’. Public Relations Review 41 (Special Issue: New Digital Publics), 2015, pp. 278–87.

Goffman, Erving: Interaction Ritual: Essays in Face-to-Face Behavior. Chicago: Aldine Pub. Co., 1967.

Jackson, Nigel / Lilleker, Darren: ‘Microblogging, Constituency Service and Impression Management: UK MPs and the Use of Twitter’. The Journal of Legislative Studies 17 (1), 2011, pp. 86–105.

Larsson, Anders O. / Moe, Hallvard: ‘Twitter in Politics and Elections. Insights from Scandinavia’. In Twitter and Society, edited by Katrin Weller, Axel Bruns, Jean Burgess, and Merja Mahrt, 219–30. New York: Peter Lang Publishing Inc, 2013, pp. 319–330.←73 | 74→

Parmelee, John H. / Bichard, Shannon L.: Politics and the Twitter Revolution: How Tweets Influence the Relationship between Political Leaders and the Public. Lanham, Md.: Lexington Books, 2013.

Marie-Anne Paveau. ‘Genre de discours et technologie discursive. Tweet, twittecriture et twitterature’. 2012,, accessed on 06.09.2015.

Sæbø, Øystein: ‘Understanding Twitter Use among Parliament Representatives: A Genre Analysis’. In Electronic Participation, edited by Efthimios Tambouris, Ann Macintosh, and Hans de Bruijn, 1–12. Lecture Notes in Computer Science 6847. Springer: Berlin Heidelberg, 2011,, accessed on 17.09.2015.

Small, Tamara A. : ‘La politique canadienne en 140 caracteres : la vie des partis dans l’univers Twitter’. Revue parlementaire canadienne 33 (3), 2010, pp. 41–48.

Thimm, Caja / Einspänner, Jessica / Dang-Anh, Mark: ‘Politische Deliberation Online – Twitter Als Element Des Politischen Diskurses’. In Mediatisierte Welten. Wiesbaden: VS Verlag fur Sozialwissenschaften, 2012a, pp 283–305.

Thimm, Caja / Einspänner, Jessica / Dang-Anh, Mark: ‘Twitter Als Wahlkampfmedium: Modellierung Und Analyse Politischer Social-Media-Nutzung’, Publizistik, 3 (57), 2012b, pp. 293–313.

Thimm, Caja / Einspänner, Jessica / Dang-Anh, Mark: ‘Mediatisierung Und Medialität in Social Media: Das Diskurssystem “Twitter” ’. In Sprache Und Kommunikation Im Technischen Zeitalter. Wieviel Technik (v)erträgt Unsere Gesellschaft? Berlin: de Gruyter, 2013, pp. 68–91.

Vergeer, Maurice / Hermans, Liesbeth: ‘Campaigning on Twitter: Microblogging and Online Social Networking as Campaign Tools in the 2010 General Elections in the Netherlands’. Journal of Computer-Mediated Communication 18 (4), 2013, pp. 399–419.

Vion, Robert: La Communication Verbale. Analyse Des Interactions. Paris: Hachette, 1992.

Zappavigna, Michele: Discourse of Twitter and Social Media: How We Use Language to Create Affiliation on the Web. 1st ed. Bloomsbury Academic, 2012.

Бахтин, Михаил: Собрание сочинений. Vol. 5. Москва: Русские словари, 1996.←74 | 75→

17 The use of the term “interaction” to describe a tweet sent by one person may seem inappropriate. However, we consider it justified by the idea that the person composing the tweet has in mind a potential audience: it is a social act in Goffman’s sense, obeying Bakhtinian principles of dialogism (infra).

18 Where several candidates had the same score for a particular communication operator, the first one in the list of results was selected.

19 The percentage is calculated in relation to the total number of tweets produced by a candidate during the period of analysis.

20 By “media” we understand all photos & videos that were used by a candidate in her/his tweet.

21 Because of the way in which the data was collected, it was unfortunately not possible to reproduce screenshots of all tweets showing them in their original context. For this reason, only the text of the tweets is reproduced here.

22 Screenshots of newspapers’ sites with relevant information, most often a picture of a candidate, are common. Their use in a tweet may fulfil an argumentative function (reference to an external source) or be a means of self-promotion.

23 This distinction is of course largely artificial, since even “spontaneous” photos are planned and taken intentionally. The difference here is more precisely as to how explicit the staging appears to be.

24 It should be further noted that this “potential” remains a prescriptive idealisation when based on an analysis of the dialogical features of the tweets themselves. Another way to approach the question would be to analyse the actual “impact” of the tweets in the Twittersphere, based on conversations and retweets (supra).

25 It is not possible to show separate examples for each type of operator, because the functions are actualized by the way of correlations between different operators, thus to extract them from their context does not make sense.

26 One particular limit to characterising the dialogical value of retweets is the difficulty of knowing whether a message is being quoted partially or not, and consequently distinguishing its different authors. While this is possible when working on a corpus, it is much more difficult for users in real time.