Are You Following Me? A Content Analysis of TV Networks' Brand Communication on Twitter

Jhih-Syuan Lin and Jorge Peña

University of Texas at Austin

Abstract

This study provides a content analysis of television networks' relational messages on Twitter by employing Bales's interaction process analysis method. It also explores the phenomena of information diffusion and influence through retweeting behavior. The findings show that television networks employed more task than socioemotional communication across program genres, but suggestions were the most frequently used message content. In addition, more positive than negative socioemotional messages appeared, and socioemotional messages were retweeted more often than task-oriented messages. In response to the ever-changing media marketplace, this study extends the scope of current media branding research and suggests managerial implications for networks' brand management and relationship-building efforts through social networking sites.

Keywords: Social media marketing, media branding, interaction process analysis, socioemotional and task communication, diffusion of information

Social networking sites (SNS) such as Facebook and Twitter have posted double-digit growth and are expected to reach nearly 148 million (or 63.7% of) U.S. Internet users in 2011 and 164.2 million (67%) of them by 2013 (Williamson 2011). Users are spending more time on SNS as well, growing from an average of 3 hours per week in December 2008 to more than 5.5 hours in December 2009 (Nielsenwire 2010a). The increasing popularity of SNS has brought extensive attention to the need to understand how these applications affect the nature of online conversations, interpersonal relationships, and communicative outcomes (e.g., Ellison, Steinfield, and Lampe 2007; Walther et al. 2009). Such computer-mediated communication (CMC) research is critical to the advertising and marketing field, where both academics and practitioners aim to determine the effectiveness of consumer-brand interactions in SNS and their consequential effects on online and offline consumer behavior (Bagozzi and Dholakia 2006; Brown, Broderick, and Lee 2007; McWilliam 2000; Taylor, Lewin, and Strutton 2011).

Twitter is "a real-time short messaging service that works over multiple networks and devices" (Twitter.com). Its technology characteristics, such as brevity, access mobility, and broadcast nature, set this communication application apart from other communication media (Zhao and Rosson 2009) and render it a strong force in the online ecosystem (comScore 2011). To explicate the use of this microcontent SNS, some researchers have begun to examine how and why people use Twitter (e.g., Java et al. 2007; Krishnamurthy, Gill, and Arlit 2008). For example, Kim and Lee (2010) note that college students use Twitter for six main reasons: entertainment, passing time, social interaction, information seeking, information providing, and professional advancement. Other studies suggest that Twitter serves both interpersonal and mass communication purposes (e.g., Jansen et al. 2009; Palser 2009). Twitter helps users keep in touch with friends and colleagues, gather information for work and personal interests, and broadcast messages of up to 140 characters to a large audience of "followers" (Zhao and Rosson 2009).

Twitter also provides the opportunity for dialogic communication between companies and consumers (Edman 2010; Jansen et al. 2009). Large media companies recognize the potential of SNS to help them reach a "subscribed" audience and deepen relationships with them (Carter 2009; Kang and Vranica 2007). However, we know little about how media brands use SNS to communicate with their consumers and followers. Instead, most available television branding studies examine the uses and effects of network websites as a form of brand extension (e.g., Ha and Chan-Olmsted 2004). Thus researchers have not fully investigated how traditional media brands craft online interpersonal messages (e.g., status updates, Tweets) to communicate with current and prospective consumers in SNS. For example, we do not know which types of interpersonal messages get produced by media companies on Twitter, nor which messages exert more influence or get repeated more frequently by SNS users.

In this study, we therefore analyze how television networks communicate interpersonally with viewers in SNS (e.g., Twitter) by employing a well-established content analysis method called interaction process analysis (IPA, Bales 1950). This investigation also considers the positive and negative valence of television networks' online messages across program genres and explores the phenomena of information diffusion and influence manifested by retweeting behavior (i.e., reposting a message originated by a different sender to one's own online social network). Recent studies view retweeting as a reliable indicator of the popularity and influence of specific messages (boyd, Golder, and Lotan 2010; Kwak et al. 2010), so understanding the content and influence of online messages posted by traditional media companies can clarify how these companies use SNS to influence consumers.

Theoretical Background and Literature Review

Relational Content of Online Messages

Various studies examine mediated communication processes and how individual communicators derive meaning from message content in online contexts (Naidu and Järvelä 2006; Peña and Hancock 2006). For example, prior studies investigate differences between text-based computer-mediated interactions and face-to-face (FtF) communication for work quality, productivity, and task satisfaction (e.g., Straus and McGrath 1994), socioemotional and task communication in different organizations (e.g., Rice and Love 1987), and the effects of group decision-making support software (e.g., McGrath and Hollingshead 1994).

Prior literature investigating the content of online interpersonal messages also has applied the IPA method, which offers a longstanding, influential means to study human interaction (Hirokawa 1988; McGrath 1984). Bales's (1950) IPA categorizes communication according to its purported goal, ranging from instrumental inquiries about a task at hand (i.e., task communication) to expressions of social information and emotions (i.e., socioemotional communication). In addition, IPA provides an observation system based on 12 interrelated categories: 6 task instrumental and 6 social-emotional areas (see Table 1). The resulting systematic framework has been applied to study instrumental (Rice and Love 1987) and recreational (Peña and Hancock 2006) computer-mediated interactions; thus, IPA provides an influential approach to interpersonal communication in offline and online contexts (see Walther 1992).

Table 1.  Categories of Bales's IPA Observation System

Message Orientations

Examples

 

Task/Instrumental

 

Ask for opinion

Evaluation, analysis, expression of feeling

What do you think about the program?

Ask for suggestion

Direction, possible ways of action

What can the actor do in the given situation?

Ask for information

Information, repetition, confirmation

Will she unleash a former house guest tonight?

Gives opinion

Evaluation, analysis, expresses feeling, wish

The show was really amazing tonight.

Gives suggestion

Direction, implying autonomy for other

Watch the show tonight.

Gives information

Information, repeats, clarifies, confirms

The preview of tonight's episode is now available online.





Positive Socioemotional

 

 

 

Shows solidarity

Raises other's status, gives help, reward

Thanks so much for liking the show.

Shows tension release

Jokes, laughs, shows satisfaction

Wow, that was funny.

Agree

Shows passive acceptance, understands, concurs, complies

Yeah, I agree with you.

Negative Socioemotional

 

Disagree

Shows passive rejection, formality, withholds help

I told you that's not allowed in here.

Shows tension

Asks for help, withdraws out of field

I am not happy about it.

Shows antagonism

Deflates other's status, defends of asserts self

Why don't you just shut up?





Sources: Bales (1970); Peña and Hancock (2006).

Although distributions across IPA categories appear to follow an orderly pattern with generally higher proportions of task-oriented messages, the pattern varies with several contextual features, including the task at hand (McGrath 1984). For example, therapy groups may incorporate more positive and negative socioemotional than task communication; decision-making teams communicate using more task-oriented messages. The distinction between task and socioemotional communication thus underlies several historically imperative distinctions of groups (McGrath 1984).

Applied to this study context, it is reasonable to expect that television networks employ task (e.g., "Watch the latest episode") more frequently than socioemotional (e.g., "Hey, good to know you enjoyed the show") communication, because the former serves to coordinate actions and provide information (Bales 1950, 1953, 1970; McGrath 1984), in line with television networks' intentions to promote on-air programs and transmit program-related information to viewers through SNS. Moreover, networks aim to publicize and steer traffic to their programs and websites to increase audience share. Therefore, it seems reasonable to assume that television network brands focus more on giving suggestions to consumers when communicating through Twitter. Accordingly, we pose two hypotheses to examine the message orientation and message content television networks use on Twitter to communicate with viewers.

H1: Television networks' program tweets contain more task than socioemotional communication.

H2: Giving suggestions is the most frequently employed task-related message content in television networks' tweets.

Yet socioemotional communication, which can express agreement, disagreement, or emotions in general (e.g., "Yes, I agree," "You are not allowed to do so") (Bales 1970), also serves a reinforcement (positive or negative) function and attempts to guide people toward or away from exhibited behaviors (Bales 1953). For example, social information processing theory (SIP, Walther 1992) posits that, given time and experience, computer-mediated environments are conducive to the exchange of social information, despite the reduced availability of nonverbal cues. Over time, people may become more socioemotional and exchange more positive socioemotional online interactions (Walther 1992).

In addition, the hyperpersonal model (Walther 1996) suggests that computer-mediated interactions provide opportunities for selective presentation, idealization, and reciprocation. The message senders express and transmit information that is more desirable for achieving their social goals in CMC environments; message receivers take such information to construct idealized images of their counterparts and the ongoing relationships, then confirm them through reciprocation (Walther 1996). Thus, considering the technical capabilities of text-based communications for impression management, television networks and their representatives likely post Twitter messages containing positive socioemotional communication (e.g., jokes, salutations) more often than negative socioemotional communication, because these message senders are interested in optimizing their self-presentation through the use of Twitter.

In support of this theory, Zhao and Rosson (2009) find that Twitter complements other SNS and increases chances to exchange social support and maintain intimacy. Edman (2010) shows that companies use Twitter as a public relations communication tool to cultivate mutually beneficial relationships with the public online. Eckler and Bolls (2011) demonstrate that consumers tend to respond more favorably to positive emotional messages than to negative or mixed ones; the sense of pleasantness facilitates a positive attitude toward the messages. Therefore, television network brands likely exchange more positive than negative socioemotional tweets to reinforce consumers' choices, elicit favorable attitudes, develop long-lasting relationships, and present the programs and networks more positively. In turn, we formulate our next hypothesis to determine the socioemotional perspective of program-affiliated tweets.

H3: Television networks' tweets contain more positive than negative socioemotional communication.

Akin to marketing communication for conventional brands, media companies address their audience as consumers (Siegert, Gerth, and Rademacher 2011). Media brands' strategic decisions are driven by the uniqueness of media products (Chan-Olmested 2006; McDowell 2006), which contribute to the brand positioning and help define their promotion (Siegert, Gerth, and Rademacher 2011). Because media companies and product manufacturers use SNS strategically (Chan-Olmested 2011), we posit that program genres have critical influences on networks' online message orientations (Bielby and Bielby 1994; Gitlin 1983). In addition, prior research indicates that genres effectively typify and categorize the range of content and corresponding viewer groups (Cohen 2002; Waterman 2006). For television networks, genres reveal individual preferences and help them predict viewing patterns, so they likely tailor different types of promotional messages to fit what is prototypical of a genre to appeal to particular segments of a market (Prag and Casavant 1994). In this sense, program genres may affect the message orientations that television networks choose to communicate with viewers. Previous studies have not explored how message orientations differ across program genres using IPA categories, so we pose a research question:

RQ1: Which message orientation-task or socioemotional communication-appears more often in genre-specific program tweets?

Retweeting and Social Influence

Researchers note that one way to assess influence is to determine the extent to which people repeat ideas, follow a trend, or pay attention to the information provided by key members of a community (Phelps et al. 2004). A classic approach that relies on this assumption is diffusion of innovation theory (Rogers 1995), which focuses on how individuals and communities adopt ideas and technologies. In this process, influentials, or people who can sway others, are important because they can accelerate or retard the adoption of an innovation or technology within their communities (Roch 2005).

In marketing literature, social networks play an imperative role in the diffusion of information (Steyer, Garcia-Bardidia, and Quester 2006). This phenomenon is even more evident for the effects of online text-based communication. As previous research suggests, online text-based communication enables people to influence more others, minimizes the effort required to exert influence attempts, and increases flexibility for incorporating influence strategies; thus, influence in online social networks is more compelling and pervasive than are FtF interactions (Subramani and Rajagopalan 2003).

For example, Twitter users may follow others or be followed by other users (Twitter.com). Subscribers tend to believe that information gathered from people they elected to follow is trustworthy and useful, because they were selected on the basis of similar, shared interests (Zhao and Rosson 2009). In addition to acquiring information directly from trusted sources, Twitter users may receive messages through retweets. Recent research characterizes retweeting as an effective tool that allows users to relay information beyond their adjacent connections (e.g., boyd, Golder, and Lotan 2010; Kwak et al. 2010; Suh et al. 2010). The practice of retweeting expands the scale and scope of influence and empowers users, in that they can dictate which information is useful and timely and should be forwarded as a retweet (boyd, Golder, and Lotan 2010; Kwak et al. 2010).

The tweeting-retweeting communicative process highlights the importance of the original tweet and suggests an insightful perspective on knowledge sharing and information transfer. Retweets are driven by the content value of a tweet (Cha et al. 2010), though boyd, Golder, and Lotan (2010) also report that Twitter users prefer retweets of time-sensitive material and breaking news. Similarly, Cha and colleagues (2010) reveal that mainstream news organizations spawn more retweets over a wide range of topics. Suh and colleagues (2010) observe that some content features, such as URLs and hashtags, have strong retweet potential. According to Naveed and associates' (2011) analysis of the content-based features of retweets, tweets expressing sentiments (positive and negative values) are more likely to be retweeted, whereas including a positive emoticon lowers this probability. Tweets that end in an exclamation mark or contain direct messages are unlikely to be retweeted; those ending in a question mark are more likely.

Kwak and colleagues (2010) also examine how far and deep retweets travel and conclude that a message generally gets retweeted nearly instantly after the first retweet, signifying a fast information diffusion process. Tweets repeated frequently and spread broadly to a large number of recipients thus can be more influential (Kwak et al. 2010; Phelps et al. 2004; Suh et al. 2010). Consider an example: Terry Moran, an ABC news reporter, tweeted President Obama's off-the-record comment during a CNBC interview about musician Kanye West's outburst at the 2009 MTV VMA Awards, calling the musician a "jackass." Although Moran soon removed the tweet, millions of other Twitter users retweeted his comment in a matter of seconds and made it one of the most influential tweets (Gavin 2009).

Although extant literature reports that retweeting is a naturalistic and nonintrusive measure of message influence, this mechanism raises questions about large-scale information diffusion and the types of branded messages that are most effective in cutting through the clutter. Before the launch of Twitter, marketers and advertisers already experienced the power of emotional content by creating buzz through forwarding behavior. For example, highly emotional e-mail messages are more likely to meet the standard for forwarding and get dispatched, even by infrequent senders of pass-along e-mails (Phelps et al. 2004). Lindgreen and Vanhamme (2005) suggest that emotions drive viral marketing campaigns, and Eckler and Bolls (2011) conclude that emotional tone is a critical determinant of how consumers process viral video advertising, such that positive emotional tones result in higher forwarding intentions. However, existing research has not scrutinized the types of brand communication (e.g., task or socioemotional orientation) that are more influential according to retweeting analyses. Without available studies of the interpersonal content of media brands' retweeted messages, we examine this issue with a research question:

RQ2: Which television network tweets-task or socioemotional-get repeated more frequently?

Method

To address our hypotheses and research questions, we conducted a content analysis to determine the communication orientations used by television networks on Twitter. By definition, content analysis quantifies and analyzes the presence or absence of elements, based on a predetermined set of categories (Krippendorff 1980). Therefore, the unit of analysis was each individual tweet posted by a network's official program account (e.g., TrueBloodHBO).

Sampling

To ensure the representativeness of the sample, we created a list of 58 top-rated primetime (7:00-10:00 p.m. CDT) television programs (both broadcast and cable networks), according to ACNielsen ratings (Nielsen.com) between September 6, 2010, and November 14, 2010. We excluded sporting events, because they do not offer regular and episodic programs. We also deleted 20 programs from the list because they did not host a Twitter account at the time of this study. Thus, we retained 38 television programs for analysis.

Regarding the selection of program genres, we employed the top genres that fit within the scope of this study (i.e., drama, reality, and comedy; Nielsenwire 2010b). Thus we categorized our list of 38 programs into three genres, according to the program description listed on each program's official website: 25 drama, 7 reality, and 6 comedy series, which produced 11,034 unique tweets. Next, we randomly selected 3 programs from each genre and 150 tweets from each program, using a random number generator. Therefore, the final sample consisted of 1,350 unique tweets (450 tweets for each genre), posted by 9 television programs (see Table 2). This sample size is sufficient, according to Neuendorf's (2002) criteria. The sampling process included printouts of the networks' tweets every 24-hour period during the study time span. All tweets posted included a user name, to identify the author by logon name.

Table 2. Sample Programs in Sample

Genre

Show

Network

Twitter ID

Tweets

N

Drama

 

 

 

 

 

 

Covert Affairs

USA

CovertAffairs

750

150

 

True Blood

HBOM

TrueBloodHBO

3868

150

 

The Closer

TNT

TheCloserTNT

631

150

 

 

 

 

 

 

Reality

Survivor: Nicaragua

CBS

Survivor_Tweet

631

150

 

Big Brother 12

CBS

bigbrotherwatch

461

150

 

Project Runway

LIFE

ProjectRunway

1983

150

 

 

 

 

 

 

Comedy

 

 

 

 

 

 

The Big Bang Theory

CBS

BigBang_CBS

328

150

 

Office

NBC

theofficenbc

1092

150

 

Glee

FOX

GLEEonFOX

1290

150

Note: Total = 11034, N = 1350.

Coding Framework and Procedure

We analyzed each tweet on the basis of its descriptive information of tweets (i.e., Twitter ID, date and time of post, and exact tweet), as well as the following categories: (1) type of tweet (original post, retweet "RT," or @reply), (2) presence or absence of a trending topic, (3) presence or absence of a link, (4) number of retweets, and (5) a message orientation according to IPA (Bales 1950). The IPA categories are mutually exclusive (Bales 1950; McGrath 1984); this measure also has been validated previously and supports comparisons with previous work. Table 1 provides the details of the categories of task- and socioemotional-oriented interactions. In addition to these 12 well-established content categories, we included a code for unclassifiable messages; IPA has no such category (Peña and Hancock 2006).

Two coders received training and analyzed the tweets according to these predetermined categories. The overall intercoder reliability, according to their performance on a randomly chosen 20% of the total messages (i.e., 270), was good in terms of Krippendorff's alpha (Table 3). Specifically, for the most detailed level of coding (i.e., IPA categories), the alpha value reaches .91 (.89 and .85 for task and socioemotional communication, respectively).

Table 3. Intercoder Reliability for Program Tweets

Code Sheet Variables

Krippendorff's Alpha

Date of post

1

Time of post

1

Type of post

1

Trending topic

1

Link

1

# of Retweets

1

Message Orientations (IPA)

.91

Task 

.89

Socioemotional

.85

Results

Of the 1,350 tweets created by 9 programs, 888 (65.8%) were original posts, 327 (24.2%) were retweets, and 135 (10%) were replies. For drama, 249 were original posts, 88 were retweets, and 113 were replies; for reality shows, 322 were original posts, 106 were retweets, and 22 were replies; and for comedy, 317 were original posts, 133 were retweets, and none were replies. Of the tweets, 765 (56.7%) (176 drama, 302 reality, and 287 comedy) included a trending topic, and 764 (56.6%) (124 drama, 317 reality, 323 comedy) contained a link that led viewers to program-related web pages. After analyzing general descriptive aspects of the networks' program tweets across genres, we can examine the content of the messages and between-genre differences as they apply to Bales's IPA categories.

Hypotheses Testing

Regarding the message orientation employed in the program tweets, we compared the frequencies of task and socioemotional communication. The results showed that task communication (80.2%, 1,083 of 1,350) was more common than socioemotional communication (19.6%, 265) (c2 = .02, p < .001). Therefore, we found support for H1.

Among all tweets, as we predicted in H2, giving suggestions (36.9%, 498) was the most frequent message content, followed by giving information (33.8%, 456), showing solidarity (13.3%, 180), giving opinions (5.6%, 75), and tension release (4.9%, 66) (c2 = .01, p < .001). That is, the results offer great support for H2.

In addition, in H3 we proposed that program tweets would be more likely to produce positive socioemotional communication than negative versions. To test this assumption, we grouped the socioemotional messages into positive or negative categories. All the socioemotional communication messages were positive (100%, 265 of 265); the networks posted no negative messages at all. Therefore H3 received support.

To examine whether television programs employed different message orientations across genres, we considered the drama, reality, and comedy categories. The results showed that task communication remained the most frequently used message orientation for drama (62.7%, 282 of 450; c2 = 26.73, p < .001), reality (88.4%, 398; c2 = 25.61, p < .001), and comedy (89.6%, 403; c2 = 12.58, p < .01) (see Figure 1). With further analyses, we gained insight into the type of message orientations employed by the television networks across genres: For drama, solidarity was the most common (27.6%, 124), followed by giving information (26.9%, 121), and giving suggestions (22.4%, 101; c2 = .02, p < .001). For reality, giving suggestions was the most frequent message content (45.8%, 206), followed by giving information (33.6%, 151) and showing solidarity (6%, 27; c2 = 51.44, p < .001). Finally, in the comedy category, giving suggestions was employed most frequently (42.4%, 191), then giving information (40.9%, 184), and then showing solidarity (6.4%, 29; c2 = 68.81, p < .001). All the differences were statistically significant.

Figure 1. Message Orientations Across Genres

Message Orientations Across Genres

 

With comparisons by message orientation, we aimed to determine whether the different usage of these top three message orientations (i.e., giving suggestions, giving information, and showing solidarity) were statistically significant across genres. As we illustrate in Figure 2, reality shows (41.4%, 206 of 498) significantly relied on giving suggestion more often in their message orientations than did comedy (38.4%, 191 of 498) or drama (20.1%, 101 of 298) programs (c2 = .10, p < .001). In addition, comedy (40.4%, 184 of 456) tended to use more giving information actions than did the reality (33.1%, 151 of 456) and drama (26.5%, 121 of 456) categories (c2 = .09, p < .001). These findings also were statistically significant. Finally, drama (68.9%, 124 of 180) significantly outpaced comedy (16.1%, 29 of 180) and reality (15%, 27 of 180) when it came to showing solidarity (c2 = .04, p < .001).

Figure 2. IPA Categories by Genre

IPA Categories by Genre

 

Our second research question pertained to the mechanism of retweeting behavior, which empowers communication counterparts and allows for the delivery of information to far more people than the source's immediate follower(s), such that it implies an imperative perspective on information diffusion. We conducted an independent samples t-test and difference-of-proportions test to determine if task or socioemotional communication was more influential, according to the number of retweets. The results revealed that 22.4% (243 of 1,083) of task-related program tweets were retweeted, and 31.3% (83 of 265) of socioemotional tweets were retweeted. The frequencies ranged from 1 to more than 100 times for each tweet. On average, socioemotional communications were retweeted more often (task M = 2.50, SD = 8.15; socioemotional M = 3.38, SD = 12.81), and the findings were statistically significant (t(624.203) = -1.39, p < .001).

To clarify the retweeting phenomena for each program genre, we next conducted three sets of one-way analyses of variance. For drama, task communication messages were retweeted 2.02 times on average (SD = 8.90), and socioemotional messages were retweeted 2.36 times (SD = 9.11; F(1,448) = .15, p = .07). Although these socioemotional communications were retweeted more often than task communications, the result was not statistically significant. In the context of reality shows, on average, task communication messages were retweeted 1.07 times (SD = 4.28), whereas socioemotional messages were retweeted .96 times (SD = 1.98; F(1,448) = .03, p < .001). This difference was statistically significant; task-related messages produced by reality shows were more influential than socioemotional messages. Finally, for comedy, task communication messages were retweeted 6.61 times (SD = 18.74), whereas socioemotional messages were retweeted 4.78 times on average (SD = 8.40; F(1,446) = .42, p < .001); that is, task-oriented messages developed in relation to comedy programs were more likely to be retweeted than socioemotional communication, with a statistically significant difference.

Discussion and Implications

The purpose of this study has been to expand current understanding of brand communication in SNS by analyzing the interpersonal content of Twitter posts (i.e., tweets) produced by television networks. This study also has examined the type of interpersonal messages that get retweeted more frequently and that thus might be more influential. Overall, the findings show that television networks employ this newly evolved communication channel to publicize on-air programs and optimize relationships with their consumers by using distinct relational messages.

First, on Twitter, television networks posted more task-oriented than socioemotional tweets. They use SNSs to promote their on-air programs, escalate viewership, and boost popularity among viewers. A closer look at the message orientations reveals the great frequency of giving suggestions or commands, in support of the prediction that television networks employ SNS messages to coordinate activities, drive traffic to their programs, and offer program-related insights and information that retains viewers' interests. The dominance of task communication is consistent with results from a study by Hwang, McMillan, and Lee (2003) that reveals the information-driven nature of corporate websites. Media companies' brand communication through SNS thus shares similar orientations to those in online venues. In an emerging trend, television network brands develop online communication with a strong tactical focus, akin to commercial brands.

Second, we find that television networks post more positive than negative socioemotional tweets. In line with SIP theory (Walther 1992) and the hyperpersonal model (Walther 1996), television networks attempt to create more positive socioemotional messages for their SNS messages to optimize their self-presentation and develop more positive relationships with viewers. From a relational perspective, these findings replicate prior CMC research (e.g., Tidwell and Walther 2002; Walther et al. 2009) and theoretically confirm that networks and followers visibly exchange social resources in the context of SNS. Twitter has been employed as a public relations force to develop and maintain relationships with consumers (Edman 2010); this study provides supportive evidence that network brands manage their socioemotional communication using a positive tone. This tendency is intuitively logical, because positive emotional messages generally help brands generate positive attitudes among consumers and lead to more favorable branding outcomes (Eckler and Bolls 2011).

Third, this study indicates the types of messages that followers consider more important and influential. Socioemotional communication got retweeted more often than task-oriented communication in our sample of messages. It thus appears that expressive messages are more appealing to television networks' followers, which caused them to be forwarded more often and exert more influence than instrumental messages. Accordingly, the effect of positive emotional messages by which they increase pass-along intentions in viral marketing (e.g., Eckler and Bolls 2011; Lindgreen and Vanhamme 2005; Phelps et al. 2004) may arise for SNS messages too. Online messages that consumers find pleasant, rather than unpleasant, evoke forwarding intentions. However, the distribution of retweets between task and socioemotional communication exhibit significant variation in each program genre (cf. drama). That is, drama contexts featured more retweeted socioemotional messages, but task messages were more commonly retweeted for the reality shows and comedy. These findings offer empirical evidence of the notion that the strategic use of SNS differs, depending on the nature of the brand; some media products, such as drama, appear more compatible with the value of SNS and consumer engagement (Chan-Olmsted 2011).

Collectively, these preliminary results establish the mechanism of retweeting as an indicator of popularity and influence, in line with diffusion of innovation theory and existing empirical studies. Because retweeting disseminates information on a large scale, from influentials to additional recipients, Twitter could be an important means to spread the word and stimulate trial, adoption, or use of branded services. Some researchers therefore conceptualize Twitter as a new form of electronic word-of-mouth communication, with potential for fostering brand relationships and gauging marketplace reactions (e.g., Jansen et al. 2009; Kwak et al. 2010). In particular, electronic consumer-to-consumer communication through Twitter may be effective in increasing brand knowledge, awareness, and attitude (Brown, Broderick, and Lee 2007; Phelps et al. 2004). The potential of retweeting as a means to reach a broad set of potential consumers efficiently also is attracting growing research attention; the effects of influencer and recipient characteristics thus need to be examined more closely.

Managerial Implications

The practical implications of these findings for television networks' creation of SNS messages are extensive. Leaders in the media industry agree that social media will drive future mainstream media (Nutley 2010), and television networks' presence on SNS is well established (Carter 2009). In this extension of existing media branding literature, we suggest that television network brands should focus on developing instrumental messages, to give their content added depth and draw potential consumers to other program-related online venues, such as network websites and blogs. Our findings echo Carter's (2009) suggestions, in that such task-oriented communication is inevitable and considered a routine communication practice to reach consumers who have drifted away from traditional media. Considering the many consumers who use SNS, television networks' social media marketing requires more thorough planning; they should not consider it just an add-on to their existing televised content (Nutley 2010).

In addition, SNS characteristics provide television networks with great brand management and relationship development opportunities. Through the use of socioemotional messages, television network brands can facilitate optimized self-presentation, favorable conversations about the brands, and valuable relationships with consumers at both individual and collective levels. Modern viewers go online for their communication experiences (Carter 2009), so a television network brand might engage them in one-on-one conversations with characters or the content creation crew to encourage relationships and cement loyalty. Television networks also might incorporate feedback mechanisms to allow consumers to express their feelings about certain media content (Chan-Olmsted 2011). The economic advantages usually generated from strong brand relationships (Fournier 1998; Lemon, Rust, and Zeithaml 2001) imply that it is important for television network brands to determine not only SNS distribution strength but also the degree and type of involvement that consumers experience through SNS.

Furthermore, the dynamic, interactive online environment has transformed consumers from passive recipients to active participants in brand building (Chan-Olmsted 2011); in turn, companies should to format and distribute SNS messages that recipients consider important and that trigger message diffusion behavior. Our study findings point to a managerial gap in existing brand communication: The task-related communication emphasized by the networks often is less influential. Instead, television network brands are more likely to benefit from their access to consumers' influence networks if they offer socioemotional communication. Perhaps networks should employ more messages that contain strong socioemotional content to leverage the power of interpersonal networks and maximize opportunities for seeding buzz through retweets. Finally, it will be valuable for television networks to identify those influentials (Cha et al. 2010; Kwak et al. 2010) who retweet program-related messages and encourage them to become more influential market members (Phelps et al. 2004). Considering the complex nature of media branding and the early stage that still marks the use of SNS for marketing, the challenges marketers face are numerous, but the opportunities they can seize are nearly endless.

Limitations and Further Research

Although the findings of this study add to the body of knowledge about online relational messages, it is not without limitations. Considering the exploratory nature of this study, the effect of program genres on television networks' use of message orientations appears scattered. Additional research might look into channel effects to exclude possible confounding influences. That is, programs aired by the same television network may apply the same message orientations for their online messages, regardless of the program genre. Furthermore, the highest number of retweets we obtained in this study was just over 100, so our exploration is limited in this sense. Instead of the backtracking method we used, a more systematic archive process or accurate number-tracking approach would be interesting. It also would be beneficial for follow-up research to incorporate the number of additional recipients of each tweet to gain a more thorough understanding of the extent to which companies' tweets were influential through retweets (Kwak et al. 2010). In summary, this study strives to explore how television networks employ SNS to communicate with viewers and gauge and shape buzz for branding and relationship building, but the interplay among genres, message orientations, and the degree to which the message appears important needs to be determined further.

Conclusion

This study examines how "old media" (i.e., television networks) communicate with their consumers through "new media" (i.e., Twitter). In doing so, we have detailed the type of interpersonal messages employed by television networks to maintain and attract viewers (e.g., providing concrete information, suggestions, commands). We also clarify the types of interpersonal messages that get retweeted with more frequency (social messages, including jokes and salutations). Overall, it appears that television network brands strategically employ Twitter messages to inform their viewers, direct their attention to upcoming shows and events, and generate buzz about specific televised events. However, more frequently retweeted messages rely on a more positive social tone. Therefore, though television networks' online communication works mainly to inform and give suggestions to viewers, Twitter users are more influenced by networks' socioemotional messages.

References

Bagozzi, Richard P. and Utpal M. Dholakia (2006), "Antecedents and Purchase Consequences of Consumer Participation in Small Group Brand Communities," International Journal of Research in Marketing, 23 (1), 45-61.

Bales, Robert F. (1950), Interaction Process Analysis: A Method for the Study of Small Groups. Cambridge, MA: Addison-Wesley.

--- (1953), "The Equilibrium Problem in Small Groups," in Working Papers in the Theory of Action, Talcott Parsons, Robert F. Bales, and Edward A. Shils, eds., Glencoe, IL: The Free Press, 111-61.

--- (1970), Personality and Interpersonal Behavior. New York: Holt. Rinehart & Winston.

Bielby, William T. and Denise D. Bielby (1994), "‘All Hits Are Flukes': Institutionalized Decision Making and the Rhetoric of Network Prime-Time Program Development," American Journal of Sociology, 99 (5), 1287-313.

boyd, dannah, Scott Golder, and Gilad Lotan (2010), "Tweet, Tweet, Retweet: Conversational Aspects of Retweeting on Twitter," in Proceedings of the 43rd Hawaii International Conference on System Sciences. IEEE, Kauai, HI.

Brown, Jo, Amanda J. Broderick, and Nick Lee (2007), "Word of Mouth Communication within Online Communities: Conceptualizing the Online Social Network," Journal of Interactive Marketing, 21(3), 2-20.

Carter, Meg (2009), "TV and Social Media: Fighting for Dominance," New Media Age, April 16, 16-17.

Cha, Meeyoung, Hamed Haddadi, Fabrício Benevenuto, and Krishna P. Gummadi (2010), "Measuring User Influence in Twitter: The Million Follower Fallacy," in Proceedings of the 4th International AAAI Conference on Weblogs and Social Media, Association for the Advancement of Artificial Intelligence, 10-17.

Chan-Olmsted, Sylvia (2006), "Issue in Media Management and Technology," in Handbook of Media Management and Economics, Alan B. Albarran, Sylvia M. Chan-Olmsted, and Michael O. Wirth, eds. Mahwah, NJ: Lawrence Erlbaum Associates, 251-73.

--- (2011), "Media Branding in a Changing World: Challenges and Opportunities 2.0," International Journal on Media Management, 13 (1), 3-19.

Cohen, Jonathan (2002), "Television Viewing Preferences: Programs, Schedules, and the Structure of Viewing Choices Made by Israeli Adults," Journal of Broadcasting & Electronic Media, 46 (2), 204-21.

comScore (2011), "comScore Media Metrix Ranks Top 50 U.S. Web Properties for February 2011," http://www.comscore.com/Press_Events/Press_Releases/2011/3/comScore_Media_Metrix_Ranks_Top_50_U.S._Web_Properties_for_February_2011.

Eckler, Petya and Paul Bolls (2011), "Spreading the Virus: Emotional Tone of Viral Advertising and Its Effect on Forwarding Intentions and Attitudes," Journal of Interactive Advertising, 11 (2), http://jiad.org/article142.

Edman, Haley (2010), "Twittering to the Top: A Content Analysis of Corporate Tweets to Measure Organization-Public Relationships," master's thesis, Louisiana State University.

Ellison, Nicole B., Charles Steinfield, and Cliff Lampe (2007), "The Benefits of Facebook ‘Friends': Social Capital and College Students' Use of Online Social Network Sites," Journal of Computer-Mediated Communication, 12 (4), 1143-68.

Fournier, Susan (1998), "Consumers and Their Brands: Developing Relationship Theory in Consumer Research," Journal of Consumer Research, 24 (4), 343-73.

Gavin, Patrick (2009), "Obama Calls Kanye ‘Jackass,'" Politico, September 15, http://www.politico.com/click/stories/0909/did_obama_call_kanye_a_jackass.html.

Gitlin, Todd (1983), Inside Prime Time. New York: Pantheon Book.

Ha, Louisa, and Sylvia M. Chan-Olmsted (2004), "Cross-Media Use in Electronic Media: The Role of Cable Television Web Sites in Cable Television Network Branding and Viewership," Journal of Broadcasting & Electronic Media, 48 (4), 620-45.

Hirokawa, Randy Y. (1988). "Group Communication Research: Considerations for the Use of Interaction Analysis," in A Handbook for the Study of Human Communication: Methods and Instruments for Observing, Measuring, and Assessing Communication Process, Charles H. Tardy, ed. Norwood, NJ: Ablex, 229-45.

Hwang, Jang-Sun, Sally J. McMillan, and Guiohk Lee (2003), "Corporate Web Sites as Advertising: An Analysis of Function, Audience, and Message Strategy," Journal of Interactive Advertising, 3 (2), http://jiad.org/article32.

Jansen, Bernard J., Mimi Zhang, Kate Sobel, and Abdur Chowdhury (2009), "Twitter Power: Tweets as Electronic Word of Mouth," Journal of the American Society for Information Science and Technology, 60 (11), 2169-88.

Java, Akshay, Xiaodan Song, Tim Finin, and Belle Tseng (2007), "Why We Twitter: Understanding Microblogging Usage and Communities," in Proceedings of the 9th WebKDD and 1st SNA-KDD Workshop on Web Mining and Social Network Analysis, ACM Press, San Jose, CA, 56-65.

Kang, Stephanie, and Suzanne Vranica (2007), "Networks Try ‘Twittering' to Spread Their Message," The Wall Street Journal, July 16, http://online.wsj.com/public/article/SB118453646990566995-ZxuimOYJ_dieGk2KbTsW_zFUNQg_20080715.html?mod=rss_free.

Kim, Mijung and Mira Lee (2010), "Why Do College Students Use Twitter?" Association for Education in Journalism and Mass Communication Annual Conference, Denver, CO.

Krippendorff, Klaus (1980), Content Analysis: An Introduction to Its Methodology. Newbury Park, CA: Sage Publication.

Krishnamurthy, Balachander, Phillipa Gill, and Martin Arlitt (2008), "A Few Chirps about Twitter," ACM SIGCOMM Conference, Seattle, WA.

Kwak, Haewoon, Changhyun Lee, Hosung Park, and Sue Moon (2010), "What Is Twitter, a Social Network or a News Media?" 19th International World Wide Web (WWW) Conference, Raleigh, NC.

Lemon, Katherine N., Roland T. Rust, and Valarie A. Zeithaml (2001), "What Drives Customer Equity?" Marketing Management, 10 (1), 20-25.

Lindgreen, Adam and Joelle Vanhamme (2005), "Viral Marketing: The Use of Surprise," in Advances in Electronic Marketing, Irvine Clarke and Theresa B. Flaherty, eds. Hershey, PA: Idea Group Publishing, 122-38.

McDowell, Walter S. (2006), "Issue in Marketing and Branding," in Handbook of Media Management and Economics, Alan B. Albarran, Sylvia M. Chan-Olmsted, and Michael O. Wirth, eds. Mahwah, NJ: Lawrence Erlbaum Associates, 229-50.

McGrath, Joseph E. (1984), Group: Interaction and Performance. Englewood Cliffs, NJ: Prentice-Hall.

--- and Andrea B. Hollingshead (1994), Group Interacting with Technology: Ideas, Evidence, Issues, and an Agenda. Thousand Oaks, CA: Sage.

McWilliam, Gil (2000), "Building Stronger Brands through Online Communities," Sloan Management Review, 41 (3), 43-54.

Naidu, Som and Sanna Järvelä (2006), "Analyzing CMC Content for What?" Computers and Education, 46 (1), 96-103.

Naveed, Nasir, Thomas Gottron, Jérôme Kunegis, and Arifah Che Alhadi (2011), "Bad News Travels Fast: A Content-Based Analysis of Interestingness on Twitter," in Proceedings of the 3rd International Conference on Web Science, Koblenx, Germany, 1-7.

Neuendorf, Kimberly A. (2002), The Content Analysis Guidebook. Thousand Oaks, CA: Sage Publications.

Nielsen.com (2010), News & Insights/Top Tens & Trends/Television, http://www.nielsen.com/us/en.html.

Nielsenwire (2010a), "Led by Facebook, Twitter, Global Time Spent on Social Media Sites up 82% Year over Year," http://blog.nielsen.com/nielsenwire/global/led-by-facebook-twitter-global-time-spent-on-social-media-sites-up-82-year-over-year/.

--- (2010b), "Drama Leading Prime Time Genre," http://blog.nielsen.com/nielsenwire/media_entertainment/snapshot-of-u-s-television-usage-what-we-watch-and-how/.

Nutley, Michael (2010), "TV Needs to Plan ahead When Using Social Media," New Media Age, April, http://www.nma.co.uk/opinion/tv-needs-to-plan-ahead-when-using-social-media/3012461.article.

Palser, Barb (2009), "Hitting the Tweet Spot," American Journalism Review, 31 (April/May), 54.

Peña, Jorge, and Jeffrey Hancock (2006), "An Analysis of Socioemotional and Task Communication in Online Multiplayer Video Games," Communication Research, 33 (1), 92-109.

Phelps, Joseph E., Regina Lewis, Lynne Mobilio, David Perry, and Niranjan Raman (2004), "Virtual Marketing or Electronic Word-of-Mouth Advertising: Examining Consumer Responses and Motivations to Pass Along Email," Journal of Advertising Research, 44 (4), 333-48.

Prag, Jay, and James Casavant (1994), "An Empirical Study of the Determinants of Revenue and Marketing Expenditures in the Motion Picture Industry," Journal of Cultural Economics, 18 (3), 217-35.

Rice, Ronald E. and Gail Love (1987), "Electronic Emotion: Socioemotional Content in a Computer-Mediated Communication Network," Communication Research, 14 (1), 85-108.

Roch, Christine H. (2005), "The Dual Roots of Opinion Leadership," Journal of Politics, 67 (1), 110-31.

Rogers, Everett M. (1995), Diffusion of Innovations. New York: The Free Press.

Siegert, Gabriele, Matthias A. Gerth, and Patrick Rademacher (2011), "Brand Identity-Driven Decision Making by Journalists and Media Managers-The MBAC Model as a Theoretical Framework," International Journal on Media Management, 13 (1), 53-70.

Steyer, Alexandre, Renaud Garcia-Bardidia, and Pascale Quester (2006), "Online Discussion Groups as Social Networks: An Empirical Investigation of Word-of-Mouth on the Internet," Journal of Interactive Advertising, 6 (2), http://jiad.org/article80.

Straus, Susan G. and Joseph E. McGrath (1994), "Does the Medium Matter? The Interaction of Task Type and Technology on Group Performance and Member Reactions," Journal of Applied Psychology, 79 (1), 87-97.

Subramani, Mani R. and Balaji Rajagopalan (2003), "Knowledge-Sharing and Influence in Online Social Networks," Communications of the ACM, 46 (12), 300-307.

Suh, Bongwon, Lichan Hong, Peter Pirolli, and Ed H. Chi (2010) "Want to be Retweeted? Large Scale Analytics on Factors Impacting Retweet in Twitter Network," in Proceedings of the 2nd IEEE International Conference on Social Computing, Minneapolis, MN, 177-84.

Taylor, David G., Jeffrey E. Lewin, and David Strutton (2011), "Friends, Fans, and Followers: Do Ads Work on Social Networks? How Gender and Age Shape Receptivity," Journal of Advertising Research, 51 (1), 258-75.

Tidwell, Lisa C., and Joseph B. Walther (2002), "Computer-Mediated Communication Effects on Disclosure, Impressions, and Interpersonal Evaluations: Getting to Know One another One Bit at a Time," Human Communication Research, 28 (3), 317-48.

Twitter.com (2010), Help Center/Twitter Basics, http://twitter.com.

Walther, Joseph B. (1992), "Relational Communication in Computer-Mediated Interaction," Human Communication Research, 19 (1), 50-88.

--- (1996), "Computer-Mediated Communication: Impersonal, Interpersonal, and Hyperpersonal Interaction," Communication Research, 23 (1), 3-43.

---, Brandon Van Der Heide, Lauren M. Hamel, and Hillary C. Shulman (2009), "Self-Generated versus Other-Generated Statements and Impressions in Computer-Mediated Communication: A Test of Warranting Theory Using Facebook," Communication Research, 36 (2), 229-53.

Waterman, David (2006), "The Economics of Media Programming," in Handbook of Media Management and Economics, Alan B. Albarran, Sylvia M. Chan-Olmsted, and Michael O. Wirth, eds. Mahwah, NJ: Lawrence Erlbaum Associates, 387-416.

Williamson, Debra A. (2011), "US Social Network Usage: 2011 Demographic and Behavioral Trends," eMarketer, http://www.emarketer.com/Reports/All/Emarketer_2000777.aspx.

Zhao, Dejin, and Mary B. Rosson (2009), "How and Why People Twitter: The Role that Micro-Blogging Plays in Informal Communication at Work," ACM International Conference on Supporting Group Work, Sanibel Island, FL.

About the Authors

Jhih-Syuan Lin (M.A., University of Florida) is a doctoral student in the Department of Advertising, University of Texas at Austin. Her research interests include media psychology, consumer-brand relationships, interactive advertising, and cross-cultural consumer psychology. E-mail: ElaineLin@utexas.edu

Jorge Peña (Ph.D. 2007, Cornell University) is an assistant professor in the Department of Communication Studies, University of Texas at Austin. His research focuses on cognitive, affective, and behavioral processes involved in online collaboration and play. E-mail: jorge.pena@mail.utexas.edu