Consumer Tweets about Brands: A Content Analysis of Sentiment Tweets about Goods and Services

Article Consumer Tweets about Brands: A Content Analysis of Sentiment Tweets about Goods and Services Journal of Creative Communications 10(2) 176–1...
Author: Evelyn Floyd
1 downloads 0 Views 247KB Size
Article

Consumer Tweets about Brands: A Content Analysis of Sentiment Tweets about Goods and Services

Journal of Creative Communications 10(2) 176–185 © 2015 Mudra Institute of Communications SAGE Publications sagepub.in/home.nav DOI: 10.1177/0973258615597406 http://crc.sagepub.com

Jos Hornikx1 Berna Hendriks2 Abstract Social media allow consumers to easily share positive or negative information about a brand with other consumers, for instance, through Twitter. Such Twitter use is a source of information that may affect the brand reputation. Therefore, it is important to gain more understanding of how Twitter is employed to evaluate brands and to communicate these evaluations with others. Previous research on Twitter use has shown that tweets about brands are more likely to be positive than negative. The present study integrates an agenda-setting perspective with studies on word of mouth and services marketing, which have suggested that this finding may be different for services than for goods. A quantitative content analysis of 1,920 Dutch tweets for 24 different brands was performed. The analysis showed that services received significantly more negative sentiment tweets than products. Implications of these results for monitoring consumers are discussed. Keywords Twitter, reputation, word of mouth, services, negativity bias

Introduction For businesses, a strong reputation is essential for hiring qualified employees, attracting investors and persuading customers to buy goods and services (Fombrun, 1996). A reputation is based on the perceptions that stakeholders have about the businesses’ achievements, current activities and future policy (Walker, 2010). These perceptions originate from stakeholders’ direct experience with the company, purchase behaviour, advertising and news reports. News is indeed one of the more important drivers of reputation perceptions (for example, Carroll & McCombs, 2003; Meijer & Kleinnijenhuis, 2006). News may have the form of reports in newspaper and magazines, but can also occur in the form of shorter messages distributed through social media. Research has underlined the importance of social

1 2

Associate Professor, Centre for Language Studies, Radboud University, Nijmegen, the Netherlands. Assistant Professor, Centre for Language Studies, Radboud University, Nijmegen, the Netherlands.

Corresponding author: Jos Hornikx, Centre for Language Studies, Radboud University, Erasmusplein 1, 6525 HT, Nijmegen, the Netherlands. E-mail: [email protected]

Hornikx and Hendriks 177 media for business communication, such as, through blogging (Yang & Lim, 2009), Facebook (Lillqvist & Louhiala-Salminen (2014) and Twitter (Hwang, 2012). Twitter, in particular, has been put forward as a communication means to stimulate dialogue between businesses and consumers, even though businesses mainly seem to use Twitter for one-way messages (Waters & Jamal, 2011). However, for businesses, it is essential to monitor consumers’ opinions on Twitter to be able to accurately react to complaints and to foresee potential reputational harm. As Macnamara and Zerfass (2012, p. 299) put it, ‘content analysis…should be undertaken [by companies] to identify the issues and topics being discussed, source quoted, and the tone of content—that is, whether it is positive or negative for the organization’. However, their survey showed that most companies do not yet undertake such activities. In this article, such a content analysis is conducted. While research on brand-related Twitter use has shown that tweets about organizations are more likely to be positive than negative (Jansen, Zhang, Sobel & Chowdury, 2009), this article extends this insight by arguing and demonstrating that this finding does not hold for services. Services are particularly subject to negative evaluations, and this effect is reflected in Twitter use. Consumers’ use of Twitter to exchange information about brands is first approached from the agendasetting perspective that information may affect the reputation of brands. Next, we discuss the valence of tweets in light of the negativity bias principle and research on word of mouth (WOM). Consequently, we base our expectations about potential differences between tweets about goods and services on the perspective of WOM and services marketing.

Consumers Exchange Information about Brands Knowledge of the benefits of a strong reputation for companies (Deephouse, 2000; Fombrun, 1996; Walker, 2010) has been an impetus for research investigating what factors may affect this reputation. One of the important factors determining a reputation is the information that consumers receive about these brands. This information may result from consumers’ direct contacts with brands but, more typically, originates in information/reviews offered by other consumers or in indirect reports in newspapers, magazines, radio and televised news. The idea that news may affect reputation is based on the agendasetting theory, which argues that issues that figure prominently in the news are also the issues that are important to the public (McCombs & Shaw, 1972). Years of research, usually in the field of political communication, have amply demonstrated that there exist strong, positive correlations between issues on the media agenda and issues on the public agenda (Wanta & Ghanem, 2007). Carroll and McCombs (2003) argue that, similar to the political arena, the relationship between the media agenda and the public agenda can also be found in the business domain, and that the number of positive issues in the news covariates with a strong reputation. Empirical studies have indeed shown these relationships between media agenda, public agenda and reputation (for example, Einwiller, Carroll & Korn, 2010; Meijer & Kleinnijenhuis, 2006). Meijer and Kleinnijenhuis (2006), for instance, performed a content analysis of business issues that participants could have read in the newspapers or viewed on news broadcasts. They combined this media agenda with public agenda data (issues that participants associated with the companies) and participants’ reputation score of the company. Data were in line with agenda-setting predictions: companies received a higher reputation when they were associated with issues that they positively deal with, and which had prominently occurred in the news. Next to these news reports, social media present a new source of information for consumers to assess a brand’s reputation. Although traditional news media may certainly be regarded as more powerful

178

Journal of Creative Communications 10(2)

institutions to bring about agenda setting than social media, these social media also present information about brands. In addition, social media platforms, such as Twitter, allow consumers themselves to express and share their evaluative opinions about businesses and organizations (Alexandrov, Lilly & Babakus, 2013; Argenti, 2006; El Ouirdi, El Ouirdi, Segers & Henderickx, 2014; Muntinga, Moorman & Smit, 2011). Jansen et al. (2009) consider tweets as electronic word of mouth (eWOM), a form of WOM, which is product advice that consumers give to each other. WOM and eWOM are important sources of information for consumers to form an opinion about businesses and their products (for example, de Matos & Rossi, 2008; Herr, Kardes & Kim, 1991; Keller, 2007; Lee & Youn, 2009). Kietzmann, Hermkens, McCarthy and Silvestre (2011) argue that is it essential for businesses to understand the social media landscape in order to follow customers’ feedback and complaints, and to react wisely to questions and negative comments. Argenti (2006) also underlines that technology has empowered consumers to share information, but has also enabled businesses to precisely monitor how consumers think about them. However, it seems most businesses are not yet very active in monitoring consumers. In the survey of Macnamara and Zerfass (2012), among about 800 corporate communication and public relations (PR) practitioners from Europe and Australasia, only a minority (about 25 per cent) reported having tools to monitor how brands are mentioned on social media. This article aims to contribute to a better understanding of Twitter as one of the most prominently used social media (Macnamara & Zerfass, 2012). Specifically, it aims to analyze consumers’ Twitter behaviour, because tweets have the potential to affect the reputation of a business. There are hardly any empirical studies on consumers’ Twitter use in an organizational setting. An exception is the study conducted by Jansen et al. (2009). They demonstrated that in an American sample of about 150,000 tweets, 19 per cent mentioned an organizational brand. This means that Twitter is indeed used frequently in relation to brands. Not all brand-related tweets provide evaluative information about brands. The category of tweets that does provide this information is called sentiment tweets, ‘the expression of opinion concerning a brand, including company, product, or service’ (Jansen et al., 2009, p. 2179). Because of their evaluative character, these sentiment tweets can be regarded as the most important information about brands that other consumers may use as input for a brand’s reputation. In their content analysis of 50 United States (US) brands, Jansen et al. (2009) further analyzed to what extent the tweets expressed sentiments. Only 20 per cent of all tweets about brands were found to be sentiments. Jansen et al. (2009) used automatic coding software to analyze all 150,000 tweets. They also manually coded 125 tweets, compared the manual analysis with the computerized analysis of these tweets and reported that these codings were not statistically different (in fact, there was a perfect match). This report cannot be taken as full proof for the computerized approach: the intercoder reliability of the two independent, manual coders (l = 0.85) indicates lower reliability than between these coders and the computer, and the number of observations is rather low (N = 125). In addition, one may have doubts on whether computerized software is already able to understand language in the way human coders can. For instance, irony is a complex linguistic phenomenon for which a number of different markers exist (for example, Burgers, Van Mulken & Schellens, 2012), such as in the form of quotation marks in ‘This is a “perfect” product’ or emoticons in ‘This is a perfect product ;-)’. It is unlikely that a computerized approach would lead to the same, reliable analyses as human coders. Therefore, it is important to replicate Jansen et al.’s pioneering study on consumer tweets about brands by human coders, and to examine what share of tweets can be labelled as sentiment tweets, which are potentially most important for a brand’s reputation: RQ1: To what extent do consumer tweets that mention brands express sentiments about brands?

Hornikx and Hendriks 179

Valence of Sentiment Tweets In the study of Jansen et al. (2009), 52 per cent of the sentiments tweets were coded as positive about the brand and 33 per cent were coded as negative about the brand. This means that tweets about brands were relatively positive. Results on WOM (see Keller, 2007; East, Hammond & Wright, 2007) are in line with these Twitter findings. For example, East et al. (2007) convincingly demonstrated in 15 studies that consumers are much more likely to be positive than negative in their WOM about products. Parallel to every single negative WOM utterance, there are also three positive ones. East et al. (2007) explain this effect by reasoning that for products on the market, there are more satisfied than dissatisfied consumers, otherwise the product would not be on the market anymore. Even if negative brand-related tweets are less frequent than their positive counterparts, they play a great role in consumers’ evaluation of brands and products because of the negativity bias principle. This principle holds that negative stimuli (such as information, products and events) are more salient than positive stimuli (see Rozin & Royzman, 2001; Skowronski & Carlston, 1989). People detect negative stimuli more quickly than positive stimuli (Dijksterhuis & Aarts, 2003), and negative stimuli have more impact in product evaluations than positive stimuli (Herr et al., 1991; Lim & Chung, 2011). Although negative brand-related tweets may be less frequent on Twitter than positive brand-related tweets (Jansen et al., 2009), there are reasons to believe that they are more salient and more influential than the positive brand-related tweets—increasing the relevance for businesses to monitor tweets. Prompted by research on (e)WOM, it is therefore important to examine the valence of sentiment tweets in the current analysis: H1: There are more positive than negative sentiment tweets about brands.

Brand-related Tweets about Goods versus Services One of consumers’ core motivations for brand-related use of social media is information exchange (for example, Alexandrov et al., 2013; Muntinga et al., 2011): consumers seek advice on goods and services to reduce risk on the one hand, and to provide information of direct experience with these goods and services on the other. This exchange of information, or WOM, has been claimed to be more relevant for services than for products (for example, Bansal & Voyer, 2000; Zeithaml, Berry & Parasuraman, 1993). Goods, such as electronics, have tangible attributes and a constant quality, which makes it relatively easy to evaluate a product at minimal risk. Services, on the other hand, such as provided by hotels, airlines or travel agencies, are relatively intangible and the quality is more heterogeneous (for example, Parasuraman, Zeithaml & Berry, 1985), increasing the risk surrounding the purchase of the service. That is why consumers are more prone to exchanging information with others about services than about products. The information that is shared about services tends to be negative, as studies in services marketing have shown (see Anderson, 1998 for mixed research evidence). And negative information is likely to be more easily shared, as Alexandrov et al. (2013) showed. Their findings demonstrated that consumers are more likely to share information with other consumers if their brand-related WOM is based on negative instead of positive evaluations. Therefore, it can be hypothesized that consumers, when they tweet about a brand, produce relatively more negative sentiment tweets for services than for products: H2: There are more negative sentiment tweets about brands for services than for goods. The research question and the two hypotheses were addressed in a content analysis of Dutch tweets about brands that were manually coded on type (comment, sentiment) and valence (positive, negative).

180

Journal of Creative Communications 10(2)

Method Material The content analysis was based on tweets that Dutch consumers had posted on their personal Twitter accounts and which figured one of the 24 brands selected for this study. These brands all offer goods or services for the Dutch consumer. The 24 brands were selected so as to constitute a representative sample of brands for the Dutch consumer: both the goods and the services group had to contain smaller (for example, Verkade) and larger companies (for example, Heineken), domestic (for example, Unox) and international firms (for example, Nivea), and had to represent a variety of sectors (for example, food, leisure and telecommunications). A pretest among the coders ensured that each brand was reliably categorized as a goods brands (for example, electronics manufacturer Philips) or as a service brand (for example, airline company KLM). The 12 goods brands belonged to the following sectors: fastmoving consumer goods (Bijenkorf, Blokker, Hema), food (C1000, Heineken, Unox, Verkade), home and garden (Intratuin, Praxis), electronics (Philips), cosmetics (Nivea) and oil (Shell). The 12 service brands belonged to the following sectors: travel (ANWB, Arke), transportation (KLM, NS), fast-moving consumer goods (Bol.com), leisure (Centerparcs, Efteling), telecommunications (KPN, T-Mobile, Ziggo), banking (Rabobank) and energy (Nuon). For each of the 24 brands, 80 tweets were selected with the systematic random sampling technique (see Neuendorf, 2002). When searching for the brands in the search option on www.twitter.com, every tenth tweet of the large list that Twitter generated was included, leading to a total of 80 tweets per brand. A tweet was only included if no other brand was present (cf. Jansen et al., 2009), if the tweet was in Dutch and if the sender of the tweet was not the brand itself. If a tweet did not comply with all three criteria, the next tweet was considered for inclusion. If this tweet was included, the tenth tweet following that tweet was considered with the same three criteria.

Procedure Ten Dutch judges were instructed to code tweets for different brands. Each brand-related tweet was coded by two independent judges on three variables: the type of tweet (sentiment, comment); the valence in case of sentiment tweets (predominantly positive, predominantly negative, neutral); and the intended receiver (another Twitter user, such as @maris456; the brand, such as @ArkeFly; no intended receiver). The intended receiver was not a variable of interest, but was included merely as a benchmark to assess the quality of the coders as intercoder reliability was supposed to be high for this variable. In a first round of analysis, all judges tried to work with the categorization of Jansen et al. (2009) in which sentiment tweets are distinguished from information-seeking, information-providing and comment tweets. As this did not generate reliable results, it was decided to use the simpler dichotomy sentiment– comment. In a sentiment tweet, consumers give an opinion about the brand, such as ‘nothing is as good as hema sausages’ in a tweet about the Hema brand. In a comment tweet, consumers do not give an opinion about the brand or have no opinion at all, such as, ‘I am going to Hema to grab a sandwich’ or ‘@ilseroos1 which Hema store?’. After the coding sessions, the reliability of the codings was computed with Cohen’s Kappa. For the type of receiver, reliability was excellent (l = 0.91); for valence, reliability was adequate (l = 0.72); for type of tweet, reliability was average (l = 0.56). Therefore, a second session was organized in which the judges together analyzed other brand-related tweets so as to better distinguish sentiments from comments.

Hornikx and Hendriks 181 For the second round of analysis, each pair of judges that had coded tweets for a given brand had to reach agreement in those cases for which disagreement had been found in the first round of analysis. The judges eventually fully agreed on all tweet codings on the three variables.

Statistical Tests A straightforward way to test the hypotheses would be a χ2-test that compares two groups of observations: goods and services brands. Such an analysis would mask potential differences between brands within a given group (it just compares the 12 goods brands together with the 12 service brands together). Therefore, we chose an approach that takes into account this within-group variability. We considered each brand as an observation with its own percentages of sentiments, and percentages of positive, negative and neutral sentiments. We used t-tests and non-parametric Mann–Whitney U-tests to answer the research question and to test the hypotheses. When the tests concerned all 24 brands, t-tests were used; when the two groups were compared, non-parametric tests were employed because the sample sizes of each group (n = 12) did not allow for t-tests.

Results Table 1 gives the number of tweets according to type of tweet, sentiment valence and brand. Table 1. Number of Tweets Depending on Type of Tweet (Comment, Sentiment), Sentiment Valence (Positive, Negative, Neutral) and Product Type (Good, Service) Sentiment Valence Product

Brand

Goods

TOTAL Nivea Philips Bijenkorf Blokker Hema C1000 Heineken Unox Verkade Intratuin Praxis Shell TOTAL Rabobank Nuon Centerparcs Efteling Bol.com KPN

Services

Sector Cosmetics Electronics FMCG FMCG FMCG Food Food Food Food Home and garden Home and garden Oil Banking Energy Leisure Leisure Online music, books Telecommunications

Comment

Sentiment

Positive

714 65 55 53 65 51 66 67 54 45 68 58 67 567 41 38 45 48 54 45

246 15 25 27 15 29 14 13 26 35 12 22 13 393 39 42 35 32 26 35

154 9 19 16 3 23 2 9 23 26 6 15 3 159 23 10 25 27 15 9

Negative

Neutral

79 13 5 1 3 3 6 5 11 1 6 0 11 1 4 0 3 0 8 1 5 1 7 0 10 0 203 31 12 4 28 4 7 3 3 2 8 3 25 1 (Table 1 continued)

182

Journal of Creative Communications 10(2)

(Table 1 continued) Sentiment Valence Product

Brand

Sector

T-Mobile Ziggo KLM NS ANWB Arke

Telecommunications Telecommunications Transportation Transportation Travel Travel

Comment

Sentiment

Positive

Negative

Neutral

40 53 58 48 57 40 1,281

40 27 22 32 23 40 639

1 8 8 5 13 15 313

38 16 12 26 7 21 282

1 3 2 1 3 4 44

TOTAL Source: Authors’ own.

Sentiment Tweets: Proportion and Valence The present content analysis was aimed at determining to what extent tweets with brand mentions are actually sentiment tweets about brands (RQ1). The analysis showed that 33.28 per cent (639/1,920) of the total number of brand-related tweets were sentiment tweets with opinions about brands. This percentage is significantly smaller than the percentage of comment tweets mentioning a brand (t(23) = 6.86, p < 0.001). This means that the majority of tweets in which brands occurred were not sentiment tweets about brands (66.72 per cent). H1 predicted more positive than negative sentiment tweets. An analysis did not provide support for this hypothesis: it showed that 48.98 per cent of the sentiment tweets were positive (313/639), that 44.13 per cent of sentiment tweets were negative (282/639) and that only 6.89 per cent of sentiment tweets were neutral (44/639). No significant difference was observed between the percentages of sentiments tweets with positive or negative valence (t(23) = 0.50, p = 0.62).

Goods versus Services H2 predicted that there are more negative sentiment tweets about brands with services than with goods. Evidence was found in support of this hypothesis (Mann–Whitney, p < 0.01). The percentage of negative sentiment tweets was higher for service brands (21.15 per cent of all service tweets) such as KLM and Rabobank than for goods brands (8.23 per cent of all product tweets) such as Heineken and Philips. An inconvenience of this analysis is that it is based on small sample sizes (12 versus 12 observations), whereas hundreds of tweets were coded. Therefore, an overall χ2-test was also conducted with single tweets as observations. The result was corroborated with this analysis (χ2 (2) = 29.72, p < 0.001).

Conclusion and Discussion Social media allow consumers to easily generate and share news about brands, thus providing consumers with the opportunity to (intentionally or not) harm a brand’s reputation. Jansen et al. (2009) underline the growing importance of these media for companies’ thinking about news and reputation: ‘It is apparent that microblogging services such as Twitter could become key applications in the attention economy’ (p. 2186). In light of the importance to monitor tweets, the current study replicated Jansen et al.’s (2009) study on brand-related Twitter use with manual coding, and with an additional interest in examining tweets about goods and tweets about services.

Hornikx and Hendriks 183 In the first place, the present content analysis of 1,920 Dutch tweets for 24 different brands showed that 33 per cent of the tweets mentioning a brand are sentiment tweets (RQ1); these are the tweets that might have a (positive or negative) impact on reputation. This percentage is somewhat higher than the 20 per cent that Jansen et al. (2009) reported in their analysis. Still, this finding means that the majority of tweets in which brands occur are not sentiment tweets about brands. One implication for businesses is that their monitoring operations of social media would be much more efficient if they were able to filter out non-sentiment tweets. To date, maybe, this can be most reliably done manually by webcare employees whose task is it to identify potentially relevant tweets, and to respond to them correctly and in line with the brand’s desired reputation. There are indications that computerized software will become better equipped to take over this sort of task—and, of course, perform much faster. Not only are there commercial initiatives (different tools can be found online when searching for ‘Twitter sentiment analysis’), there are also academics investigating how to develop learning machines to identify irony or sarcasm. In a study by Kunneman, Liebrecht, van Mulken and Van den Bosch (2015), it was shown that computerized software can be taught to distinguish sarcastic language in tweets. In the second place, the present study examined the valence of the sentiment tweets (H1). The data did not find evidence for H1 that sentiments would be more frequently positive than negative. There were as many positive (49 per cent) as negative (44 per cent) sentiment tweets about brands. Research on WOM has generally found a pattern where positive evaluations are more frequent, similar to Jansen et al. (2009), where more positive (52 per cent) than negative (33 per cent) sentiment tweets were recorded. A plausible explanation for this difference is that half of the tweets in the present study were related to brands with services. In fact, this study was also conducted to compare the use of brand-related Twitter for goods and for services. The two statistical tests showed that, in line with the prediction (H2) based on WOM and services marketing research, relatively more negative sentiment tweets occur about brands with services than with goods. This finding is important since it is inconsistent with the general positive findings in WOM (for example, East et al., 2007; Keller, 2007). Apparently, consumers use Twitter more prominently to share negative information about services than about products. For businesses that provide services, this finding underlines the importance of systematic monitoring of social media, such as Twitter. Systematic monitoring may aid businesses in identifying when and where potential harm may occur to their reputation. Next, negative tweets allow businesses to engage in dialogue with consumers, which is one of the specific characteristics of social media among the communication channels that businesses may use (see Gilpin, 2010). Webcare, preferably with a human tone of voice, may help to counter negative WOM (see van Noort & Willemsen, 2012). More research is needed to examine just how to react adequately to consumers on social media. A limitation of the present content study is that it consisted of a sample of only about 2,000 Dutch tweets. On the other hand, the sample contained a balanced set of brands (both in the goods and the services group), was manually coded by independent coders and was analyzed with statistical tests that took within-group variation into account. With its interest in the comparison between goods and services, this study contributes to our understanding of consumers’ media use, which may affect businesses’ reputation. Indeed, an important avenue for future research is to design studies that can reliably examine how reading one’s Twitter timeline, where brand-related tweets also figure, can have an impact on consumers’ reputation of a brand. Based on the current study, it can be concluded that brands with services should be aware of the relatively large portion of negative sentiment tweets. Identifying and analyzing tweets, together with other news reports from newspapers and magazines, allows companies to understand where potential harm to their reputation may be located.

184

Journal of Creative Communications 10(2)

References Alexandrov, A., Lilly, B., & Babakus, E. (2013). The effects of social- and self-motives on the intentions to share positive and negative word of mouth. Journal of the Academy of Marketing Science, 41(5), 531–546. Anderson, E.W. (1998). Customer satisfaction and word of mouth. Journal of Service Research, 1(1), 5–17. Argenti, P.A. (2006). How technology has influenced the field of corporate communication. Journal of Business and Technical Communication, 20(3), 357–370. Bansal, H.S., & Voyer, P.A. (2000). Word-of-mouth processes within a service purchase context. Journal of Service Research, 3(2), 166–177. Burgers, C., van Mulken, M., & Schellens, P.J. (2012). Verbal irony: Differences in usage across written genres. Journal of Language and Social Psychology, 31(3), 290–310. Carroll, C.E., & McCombs, M. (2003). Agenda-setting effects of business news on the public’s images and opinions about major corporations. Corporate Reputation Review, 6(1), 36–46. de Matos, C.A., & Rossi, C.A.V. (2008). Word-of-mouth communications in marketing: A meta-analytic review of the antecedents and moderators. Journal of the Academy of Marketing Science, 36(4), 578–596. Deephouse, D.L. (2000). Media reputation as a strategic resource: An integration of mass communication and resource-based theories. Journal of Management, 26(6), 1091–1112. Dijksterhuis, A., & Aarts, H. (2003). On wildebeests and humans: The preferential detection of negative stimuli. Psychological Science, 14(1), 14–18. East, R., Hammond, K., & Wright, M. (2007). The relative incidence of positive and negative word of mouth: A multi-category study. International Journal of Research in Marketing, 24(2), 175–184. Einwiller, S.A., Carroll, C.E., & Korn, K. (2010). Under what conditions do the new media influence corporate reputation? The roles of media dependency and need for orientation. Corporate Reputation Review, 12(4), 299–315. El Ouirdi, M., El Ouirdi, A., Segers, J., & Henderickx, E. (2014). Social media conceptualization and taxonomy: A Lasswellian framework. Journal of Creative Communications, 9(2), 107–126. Fombrun, C.J. (1996). Reputation: Realizing value from the corporate image. Boston: Harvard Business School Press. Gilpin, D. (2010). Organizational image construction in a fragmented online media environment. Journal of Public Relations Research, 22(3), 265–287. Herr, P.M., Kardes, F.R., & Kim, J. (1991). Effects of word of mouth and product attribute information on persuasion: An accessibility-diagnosticity perspective. Journal of Consumer Research, 17(4), 454–462. Hwang, S. (2012). The strategic use of Twitter to manage personal public relations. Public Relations Review, 38(1), 159–161. Jansen, B.J., Zhang, M., Sobel, K., & Chowdury, A. (2009). Twitter power: Tweets as electronic word of mouth. Journal of the American Society for Information Science and Technology, 60(11), 2169–2188. Keller, E. (2007). Unleashing the power of word of mouth: Creating brand advocacy to drive growth. Journal of Advertising Research, 47(4), 448–452. Kietzmann, J.H., Hermkens, K., McCarthy, I.P., & Silvestre, B.S. (2011). Social media? Get serious! Understanding the functional building blocks of social media. Business Horizons, 54(3), 241–251. Kunneman, F., Liebrecht, C., van Mulken, M., & van den Bosch, A. (2015). Signaling sarcasm: From hyperbole to hashtag. Information Processing and Management. 51(4), 500–509. doi:10.1016/j.ipm.2014.07.006 Lee, M., & Youn, S. (2009). Electronic word of mouth (eWom): How eWOM platforms influence consumer product judgement. International Journal of Advertising, 28(3), 473–499. Lillqvist, E., & Louhiala-Salminen, L. (2014). Facing Facebook: Impression management strategies in company– consumer interactions. Journal of Business and Technical Communication, 28(1), 3–30. Lim, B.C., & Chung, C.M.Y. (2011). The impact of word-of-mouth communication on attribute evaluation. Journal of Business Research, 64(1), 18–23. Macnamara, J., & Zerfass, A. (2012). Social media communication in organizations: The challenges of balancing openness, strategy, and management. International Journal of Strategic Communication, 6(4), 287–308.

Hornikx and Hendriks 185 McCombs, M.E., & Shaw, D.L. (1972). The agenda-setting function of mass media. Public Opinion Quarterly, 36(2), 176–187. Meijer, M., & Kleinnijenhuis, J. (2006). Issues news and corporate reputation: Applying the theories of agenda setting and issue ownership in the field of business communication. Journal of Communication, 56(3), 543–559. Muntinga, D.G., Moorman, M., & Smit, E.G. (2011). Introducing COBRAs: Exploring motivations for brandrelated social media use. International Journal of Advertising, 30(1), 13–46. Neuendorf, K.A. (2002). The content analysis guidebook. Thousand Oaks, CA: SAGE. Parasuraman, A., Zeithaml, V., & Berry, L.L. (1985). A conceptual model of service quality and its implications for future research. Journal of Marketing, 49(4), 41–50. Rozin, P., & Royzman, E.B. (2001). Negativity bias, negativity dominance, and contagion. Personality and Social Psychology Review, 5(4), 296–320. Skowronski, J.J., & Carlston, D.E. (1989). Negativity and extremity biases in impression formation: A review of explanations. Psychological Bulletin, 105(1), 67–85. van Noort, G., & Willemsen, L. (2012). Online damage control: The effects of proactive versus reactive webcare interventions in consumer-generated and brand-generated platforms. Journal of Interactive Marketing, 26(3), 131–140. Walker, K. (2010). A systematic review of the corporate reputation literature: Definition, measurement, and theory. Corporate Reputation Review, 12(4), 357–387. Wanta, W., & Ghanem, S. (2007). Effects of agenda setting. In R.W. Preiss, B.M. Gayle, N. Burrell & M. Allen (Eds), Mass media effects research: Advances through meta-analyses (pp. 37–51). Mahwah, NJ: Erlbaum. Waters, R.D., & Jamal, J.Y. (2011). Tweet, tweet, tweet: A content analysis of nonprofit organizations’ Twitter updates. Public Relations Review, 37(3), 321–324. Yang, S.-U., & Lim, J.S. (2009). The effects of blog-mediated public relations (BMPR) on relational trust. Journal of Public Relations Research, 21(3), 341–359. Zeithaml, V.A., Berry, L.L., & Parasuraman, A. (1993). The nature and determinants of customer expectations of service. Journal of the Academy of Marketing Science, 21(1), 1–12.

Author’s bio-sketch Jos Hornikx teaches in the Department of Communication and Information Sciences and conducts research on persuasive argumentation and multilingual advertising. He has published in outlets including Communication Monographs, Communication Yearbook and Journal of Business Communication. Berna Hendriks has published on (B)ELF and interlanguage pragmatics in outlets including Intercultural Pragmatics, IEEE Transactions on Professional Communication and Multilingua.