Aristotle Online? An Empirical Examination on What Makes Public Speeches Go Viral

(1)

Aristotle Online? An Empirical Examination on What Makes Public

Speeches Go Viral

Timea Klebercz (11181311); Master’s Thesis; Graduate School of Communication; Persuasion programme, Communication Science; Supervisor: dr. Daan Muntinga; 30th June 2017

(2)

Abstract

The ancient rhetorical theories of Aristotle are known to be persuasive. This thesis aims to find out whether ethos, pathos and logos are also applicable to the new forms of persuasion (i.e. virality). The study intents to discover what makes public speeches go “viral.” Specifically, it examines why certain speech videos gain more views than others. To do so, 60 TED talks from the TED2016 Conference were content analysed for applying Aristotle’s rules of persuasion in speeches: ethos, pathos, and logos. In addition to exploring whether and how these classical means of persuasion explain virality, novelty and surprise in speeches were examined as factors that potentially moderate the effect of ethos, pathos, and logos on virality. The findings, however, did not yield statistically significant results, thereby discarding Aristotle’s classical rules for a modern context. However, since this study was the first attempt of its kind, it is important to make further efforts in the field of public speeches and virality.

(3)

Introduction

Earned media, publicity that was not paid for, has a growing importance since social media becomes more and more sophisticated (Corcoran, 2009). There is a general change towards the direction of free media, which means that marketing is shifting towards a

dialogic form of communication, where costumers are receivers of communication, creators of the communication channel and the receivers, as well (Corcoran, 2009). The reach is the greatest when the most consumers share a piece of content, in other words when the content goes viral. Involvement and engagement with the costumers, therefore, is a key, if virality is the aim. Companies are keen to find a recipe for spreadable ads, which – if successful- significantly increase brand awareness and - in most cases – sales, too. Numbers of shares and views have undoubtedly become an important measure of success (e.g. Akpinar & Berger, 2017; Alhabash & McAlister, 2015). And it’s not only businesses who aim to get their ads to “go viral:” the public success of other types of communication, such as public speeches, can also be weighed based on how many views they have got, for TED talkers, politicians but even commencement speeches compete against each other online, as well. In this new online environment, where the audience is bigger and more difficult to engage with than ever, public speeches these days need to go viral if they want to maximise their reach and impact of their message. However, it despite an abundance of research into the virality of news (e.g. Berger & Milkman, 2012) and brand-related content (e.g. Eisingerich et al., 2015; Yuki, 2015), it remains unclear what makes public speeches go viral.

This thesis therefore sheds light on two issues. First, virality is a relatively new topic in media research but has quickly become the crown of digital media, as more and more organizations think of it as the holy grail (Akpinar & Berger, 2017). Therefore, more insights into what makes content spreadable is needed to confirm and extend the patterns

(4)

demonstrated in prior research. Second, while there are many perspectives on what

constitutes an excellent speech (e.g. Holtgraves & Lasky, 1999; Lagerwerf et al., 2015; Petek, 2012), in a specifically online context has not yet been subject to academic examination. Therefore this research aims to fill this gap and find an answer to what makes an online speech successful.

To examine this, this study makes use of the TED-platform. The TED (Technology, Entertainment and Design) website features over 2000 “talks” speech videos on a wide variety of subjects given by a wide variety of international speakers. (TED.com, 2017) The speeches on TED.com differ with regard to the number of views. Building on previous research on rhetoric and virality, this study examines what causes the differences in the number of views of TED talks.

The corner stone of speech theories is Aristotle’s text Rhetoric, which was written in the 4th century BC. The book describes the so-called modes of persuasion, rhetorical devices that are responsible for persuasion: it distinguishes ethos, pathos and logos. Ethos is the speaker’s authority and credibility (e.g. Halloran, 1982; McCroskey, 1966; McCroskey & Young, 1981), logos are the statistics, data and logic as a mean of persuasion (e.g Anton, 1997; Lucas, 1971), and ethos is emotional appeals used in the speech (e.g. Aristotle, 1946; Brinton, 1988; Rees, 1972). Since Aristotle, the science of persuasion has expanded

enormously. Nevertheless, his logos-pathos-ethos triangle of persuasion means is still a compass when examining speeches, it has been demonstrated that all three elements contribute to persuasion (e.g. Delia, 1970; Brinton, 1988; Hass, 1981). This paper will also use Aristotle’s triangle as a base, but tests it applicability in an online environment, focusing on whether they drive virality.

In addition, the paper looks into two potential boundary conditions of the impact of the three rhetorical elements on virality: novelty and surprise. Since TED talks are about

(5)

informing, educating, and inspiring people (e.g. Scotto di Carlo, 2015), novel elements might impact the relationship between rhetorics and virality (Eelen, Verlegh, & Van den Bergh, 2015). In a similar vein, surprising element is demonstrated to keep the attention of spectators (Teixeira, 2012) and is often connected to virality in the body of academic literature

(Schamari & Schaeffers, 2015). Therefore, the elements, thus the research will take it into account as a factor in the model.

The history of rhetoric research dates back to the time of Aristotle, but good public speaker skills remain outstandingly crucial in the 21st century. Successful presentations nowadays are important both in private and business life; therefore analyzing speeches and looking into what makes them more successful is not only in the interest of people who are preparing a TED talk but the results of this paper can generally be applicable for anyone who is ever planning to give a speech. Professional speakers, such as politicians or speechwriters, could profit from the findings even more. For them a successful presentation is defined by its ability to go viral today. Given the importance of social media in communication, their ability to inspire people and enforce policies hinges on their ability to deliver a viral speech. Thus, understanding what makes a speech go viral is crucial. Thirdly, it will further add to the existing knowledge of what makes online content go viral, which has a practical value for anyone who is invested in online advertisement.

Theoretical background Defining virality

When is a video more than popular, what can be called viral? And how far does it need to spread to fall into the category? Virality has been defined in several ways in literature. Content goes viral, if it spreads in a short period thus providing access to many potential viewers (Bonchi et al., 2013). Another approach is describing virality as electronic

(6)

word-of mouth (Derbaix &Vanhamme 2003, Phelps et al., 2004), which focuses on peers pass on messages to peers. Thirdly, users’ engagement with the content, such as viewing, accessing, liking/disliking, sharing, and commenting, is a measure of online ad effectiveness. (Tucker, 2011). If there is engagement, the online material has the chance to go viral. The definition “...a social information flow process where many people simultaneously forward a specific information item, over a short period of time, within their social networks, and where the message spreads beyond their own (social) network to different, often distant networks, resulting in a sharp acceleration in the number of people who are exposed to the message.” (Nahon & Hemsley, 2013, p.16). does stand out from all of them since it covers the human aspect, the speed of spreading, the reach the terms of both how many people will be exposed to it and how far the information travels. This thesis focuses on reach, how many people have been exposed to the speech video, how many people have seen it.

What makes things go viral?

Virality, is a successful marketing technique, as well; it is the vested interest of

companies to find out what causes virality because successful campaigns are tied to growth in sales numbers (Leskovec et al., 2007). However, failed viral campaigns are the norm, and there are only a few successful ones (Mills, 2012; Van der Lans and & Van Bruggen, 2011). Since it is assumed that as a virus, it has a contagious character and due to this nature, the sender has a limited capacity to influence the desired reach (Pitta’s, 2008). Studies already found a few reasons, why people share content. Identity goals, what people want to identify themselves with in front of their online community, are a crucial element of the decision for people to engage with the content or not (Dobele et al., 2007; Phelps et al., 2004; Wallsten, 2010). The consumer culture theory (CCT) (Arnould & Thompson, 2005), frames the consumer as an active co-creator of meaning, they use viral content as an extension of their

(7)

desired self. (Beverland et al., 2014). In the virality process, gatekeepers “actively participate in the information-retrieval and dissemination process in their communities.” (Moteyer-Duran, 1993, 119 p.) have bigger influence over the flow of information (Nahon & Hemsley, 2013). Moreover, people are more likely to listen to someone who they share a personal connection with, to the so called opinion leaders (Watts and Dotts, 2007), who have also a huge impact on the spreading of the message. Besides these factors, it is also important to note that forwarding a message or speech is a sense of agency, it requires human attention and sharing. Humour, topic, interest all play a role in the decision of sharing, too; as they are vital elements of human nature and self-expression. (Nahon & Hemsley, 2013).

The influence of Ethos, logos, pathos on virality

Ethos, logos and pathos are rhetorical strategies, which were first named by Aristotle (Aristotle, 1946). Ethos is concerned with the character of the speaker, logos is concerned with the reasonableness of the arguments in the speech and pathos is concerned with the emotions of the listener.

Ethos.

Ethos is the authority of the speaker, the character of the rhetor is crucial when it comes to persuasiveness (Halloran, 1982). In this research, not only the authority or former authority of the speaker is taken into account, but also characteristics which lead to credibility without any institutional power will be considered. Fluency, tempo, and pitch are associated with arousal (Burgoon, Kelley, Newton & Keeley-Dyreson, 1989), therefore besides the background of the speaker, confidence and enthusiasm will also be looked into, as these factors are contributing to credibility and ethos (Niebuhr, Voße & Brem, 2016), as credibility contributes to persuasion (Hass, 1981; Sørensen, 2013), and this thesis will investigate its effect on virality and relation to pathos and logos.

(8)

Logos.

Logos is the rational, persuasion through reasoning. It is defined in terms of message characteristics, and it depends on the form and structure how it transmits meaning. Logic affects the reasoning and thinking process. Reasoned discourse applies inductive and deductive reasoning and evidence within the context of the listener's field of susceptibility (Delia, 1970). Logos mainly use premises, examples, analogies, authorities, and statistics as evidence in the argumentation line (Delia, 1970). Reasoning is always a conscious process, therefore some, instead of using the category rational and emotional thinking, name it conscious and unconscious. Other important theories in persuasion research, the dual-mode processing, such as Petty and Cacioppo’s (1986) Elaboration Likelihood Model and the heuristic/systematic model (Chaiken et al., 1996), they can be both analogies to the conscious- unconscious and to the rational and emotional argumentations. These theories discuss when are people assessing the relevant information more thoughtfully and

systematically, in comparison to heuristic processing when the process is low-effort.

Argumentation and another kind of rational reasoning are normally more complex, therefore it requires higher effort, while emotions would require lower effort.

Pathos.

Lastly, pathos. This mean of persuasion is defined as feelings that lead to pleasure or pain, which influence human judgment and decision-making (Aristotle, 1984). Current academic work points out that emotion have a bigger impact on intention in the persuasion process. Zebregs, van den Putte, Neijens and De Graaf (2015) found that narrative evidence affects intention to a greater extent than statistical evidence. A narrative is what includes at least one character, who experiences at least one event. (De Graaf, Sanders & Hoeken, 2016). While beliefs and attitude all relate primarily to cognitive responses, both narrative evidence and intention relate more specifically to affective responses. (Green & Brock, 2000).

(9)

Therefore narratives are one of the most important tools of appealing to emotions and Using stories and narratives are the most frequent ways of appealing to emotions in speeches, too (Gallo, 2014). Moreover, a study on sympathy by Small et al (2006) shows when emotions are used (the problem in the paper was portrayed in a story using emotions), people are more likely to be persuaded and act, compared to when the problem is demonstrated by number. Thus, literature (e.g., Brinton, 1988; Leighton, 1985) is focusing on rhetoric identifies that indeed, emotions may be reasons for action and therefore are an important mode of

persuasion.

To further explore pathos, Trevathen (1992) also noted that emotional parts in the speech would be more remembered since it has an “emotional importance” and it can trigger the listener to take action. What is more, pathos connection to virality has been researched already with the finding that emotional content does increase the likelihood of things going viral. (Bakshy et al., 2011) A good example of when emotional impact is better shared the video of Susan Boyle’s performance in Britain's Got Talent show in 2009.

Based on the section above, the first hypothesis is the following:

H1a. More pathos than ethos in a speech will lead to higher numbers of views. H1b. More ethos than logos in a speech will lead to more views.

The moderating effect of novelty

To further add to the existing knowledge of the circumstances under which ethos, pathos, and logos predict speech virality, this study look into the role of novelty. People have an innate drive to seek out for new information. For instance, recent research shows that the presence of a written “new” label on product packages and in advertising is effective; the “new” sign led to a more positive product attitude, and also to a greater purchase and word of mouth intention (Eelen et al., 2015). TED talks are about informing, educating, and inspiring

(10)

people, which is not possible without giving new information or new ways to look at things, therefore the paper will investigate the effects of novelty, as well. Newness creates positive attitude, papers found; positive effects of novelty could be driven by exploration or curiosity (Hirschman, 1980). Since novelty is sought by humans fuels by their desires, it also means that they want to be identified with novelty, therefore the thesis expects that when the novelty is high, the effect of ethos on virality will cause higher numbers of views.

H2. Novelty positively moderates the relationship between means of persuasion and number of views, such that if a speech contains more novel elements, a speech with pathos will gain more views than a speech with pathos and less novel elements.

The moderating influence of surprise

Surprise is also essential because they keep the attention of the spectator. Teixeira (2012) Dobele, Lindgreen, Beverland, Vanhamme and van-Wijk (2007) also found that emotions in a video are necessary but not sufficient to make it go viral, imagination is needed in a unique or unforgettable way, so that it also catches the attention of the viewer. Surprise makes emotions stand out, if unique, unexpected presentation will be a lot more attention grabbing and loner remembered (Dobele et al., 2007; Derbaix and Pham,1991). From a psychological point of view, definition of surprise is a neutral and short lived emotion elicited by unexpected products/services/attributes, such as a schema discrepancy or expectancy disconfirmation (Meyer & Niepel, 1994). Numerous papers suggest that positive surprise enhances customers' attitudes (Rust & Oliver, 2000; Vanhamme, 2000) and positive word-of-mouth, as well (Derbaix &Vanhamme 2003). Word-of-moth is a key here since that is one of the approaches to the definition of virality. Therefore, this paper suggests that when a TED speech has more surprising elements it will modify the effect of pathos on number of views. Based on these, the research third hypothesis is the following:

(11)

H3. Surprise positively moderates the relationship between means of persuasion and number of views, such that if a speech contains more surprise elements, a speech with pathos will gain more views than a speech with pathos and less surprise elements.

Methodology Data collection

The units of analysis are TED talks in the research. TED is a nonprofit organization, which defines its mission as spreading ideas in the form of speeches that usually last eighteen minutes or less. It started as a conference in 1984, where the topics of Technology,

Entertainment and Design were covered; it became global in 2006 when they started to publish their videos on their website and on YouTube. Today, it covers a wide variety of topics and claimed to have, In November of 2012, TED Talks had one billion collective views and in June 2015, TED.com posted its 2,000th Talk (TED.com, 2017).TED talks aim to motivate and inspire and narratives are one of the most important element of pathos in a speech. TED talks generally express opinions and emotions linked to the topic of the speeches. They are supposedly rich in evaluation and indicate the speakers’ affective

response and stance, while they are engaging the audience with an informal and emotive tone. (Scotto di Carlo, 2015). They have been chosen for this study to contribute to rhetoric

research for three reasons: firstly, TED talks are widely popular; most people associate public speaking with TED talks. Secondly, all TED speeches follow the same format and the

speakers got equal training before their performance. This makes TED speeches comparable what makes them easier to measure. And lastly, from a practical point of view, the videos of the speeches are easily accessible on YouTube, iTunes and the TED website and mobile app. This made possible that a representative sample is captured for the tests.

(12)

Sample selection

A selection was made from speech videos published on TED’s website. Specifically, all the speeches from TED’s 2016 annual conference were analysed. 2016 is the closest year from which all the speeches are already published. There are regional TED events, as well, but this research used the annual, global event, which tries to appeal to a global audience, so it is the most similar to the online environment. 60 videos got selected out of approximately 70, -50 is a sufficient number to accurately represent the population of TED videos

(Krippendorff, 2004) - but we aimed slightly above that. They were randomly selected from the webpage, collected in an Excel sheet by the researcher and the also randomly assigned to the other coders. Several videos were excluded from the analysis because technically, they were not speeches. For instance, one the videos featured the speaker having an interview with the host on the stage instead of giving a public speech alone, therefore this one was excluded from the sample. In addition, videos that were either too short (under 6 minutes) or too long (over 21 minutes) were also excluded because having the same speech length as much as possible was a goal to make comparison more efficient and to work better with the metrics.

Coder Training and Reliability

In total, 59 videos of TED talks were coded. Three coders were trained to familiarize them with the guidelines of the codebook (see next section and Appendix). The coders got face to face introduction to the codebook, then a trial coding took place. The first inter-rater agreement numbers were not satisfactory, only 53,4% was the agreement. It was monitored from variable 7 to variable 22, the administrative part and general information on the speech (expect topic(v7)) were not included in this calculation and there was 100% agreement on that part. Most disagreeing happened on the variables number of 9, 8, 12 and 21: arguments, facts, stories, and surprise. These variables therefore had a special focus for the second

(13)

edition of the codebook that was more detailed based on the conclusions drawn from the inter-coder disagreements. Next, the co-coders got a personal or Skype training again with personalised suggestions. From the sample, 10% (6 videos) were compared, again from variable 7 to 22. This time the result of the agreement was 75,6%.

Independent variables Logos.

To measure logos four components were taken into account. There is no previous literature on how to operationalise logos, pathos and ethos to measure speeches in relation to virality. Therefore, the components were chosen from the theoretical background of the field. Firstly, it was measured how much argument (V8) was used by the speaker in the speech (Anton, 1997; Hample, 1986), with the answer options: 1, not really 2, moderately used 3, used a lot. The length of the speech was taken into consideration. The option ‘not really’ always meant 0 or 1 argument, option ‘moderately used’ differed according to the length: if a speech was 6-10 minutes long 2 arguments, if a speech was 10-15 minutes long 2-3 arguments and if a speech was 15-21 minutes long 2-4 arguments. In the ‘used a lot’

category if a speech was 6-10 minutes long 3 or more arguments were necessary, if a speech was 10-15 minutes long the speaker needed 4 or more arguments and if a speech was 15-21 minutes long 5 or more arguments. All type of arguments – induction, deduction and

abductions- were counted, and significantly longer arguments were counted as two

arguments. The next item to measure logos, was scientific facts (V9), which are observation that are scientifically accepted as truth. (Cambridge Dictionary, 2017) Here also the 1-to 3 scale was used and how many is ‘not really’, ‘moderately used’ and ‘used a lot’ was

measured by the same numbers as in the case of arguments. The third component of logos is statistical data (V10), numbers used in the speech that representing information. (Gallo,

(14)

2014) The 3 point scale was used here as well, with all the rules that were applicable above. Lastly, we measured analogies (V11), how many comparisons to historical or literal events were used in the speech as explanation or example. (Gallo, 2014) The scale again was from 1 to 3, as in every case for the independent variables. To conduct our analyses, new variables needed to be created. Variable logos was computed from the mean of the four initial

variables: how many arguments were in the speech, how many scientific facts were told, how many analogies were used and how many statistics were in the video. So the value of the new logos variable is between 1 and 3. (M=2.11, SD=.45)

Pathos.

The codebook questions continued on pathos, which also has four elements: narratives, metaphors, appealing to positive and negative emotions. (Gallo, 2014, Brinton, 1988, Leighton, 1985) First of all, we looked into how many stories (V12), at least one character experiences at least one event (De Graaf, Sanders, & Hoeken, 2016), were in the speech. And the metrics was the same scale (1-3), as for the logos variables. The ‘3, used a lot, 2, moderately used 1, not really used’ scale is ideal because the coders can evaluate the videos easier, and the agreement among the raters therefore is greater. Metaphors are also widely-known to have an effect on emotions (V13), they are a description of a person or object by referring to something that is considered to share similar characteristics to that person or object (Cambridge Dictionary, 2017; Gallo, 2014), and again the same 3 point scale is used. Next (V14), appealing to negative emotion was measured, while appealing to positive emotions (V15) (Nahon & Hemsley, 2013), as well. This had to be indicated by the coders on the usual scale, evaluating whether the speaker was aiming to evoke negative or positive emotions in the audience. Variable pathos then was computed as the mean of the original variables: how many narratives, how many metaphors, and how much appealing to

(15)

negative or positive emotions were present in the speech, which self-explanatory was again between one and three (M=2.14, SD=.45).

Ethos.

Ethos, in comparison to logos and pathos, was evaluated by three instead of four variables. (Halloran, 1982, Gallo, 2014) It started with measuring how much was the speaker talking about his professional background (V16). (Gallo, 2014) The other two variables were enthusiasms (V17) and confidence (18). These three variables also were measured on a 3 point scale (1, not really,2, moderately and 3, a lot), mentioning professional background was counted as the other independent variables above, but enthusiasm and confidence were judged by the coder based on fast tempo, variance, pitches and supporting body language and facial expression (Burgoon, Kelley, Newton & Keeley-Dyreson, 1989). The new variable ethos is the mean value of how much did the speaker talked about their professional background, enthusiasm and confidence (M=2.28, SD=.43)

Control variables

In the second section of the codebook (see Appendix), general information about the speech was coded. The gender of the speaker was coded (0 = female, 1 = male). Gender was used as a control variable, since gender stereotypes in communications (Berryman & Wilcox, 1980). Next, the celebrity status of the speaker was coded (1 = celebrity, 0 = no celebrity) to prevent any celebrity endorsement would have been used as a control variable but there were no celebrities in the sample. Celebrities work as opinion leaders, people feel a special

connection to them and they have more intention to share that content. (Burt, 1999) The length of the speech was also marked on the coded sheet (V5), this influenced how amount of arguments, etc. was scaled (see above). Since the videos were posted online a few months apart (V5), this information has been also used as a control variable, for the reason that an

(16)

older video has more time to collect views. The speeches got categorized into topics (V7), and it was controlled for this factor, as well. 1,Technology 2, Entertainment 3, Business 4, Science 5, Environment 6, Minority rights 7, International relations, role of state 8,

Motivation 9, Art 10, Lifestyle 11, Other. In the second edition of the codebook and variable another control variable, humour (V24) was added, measured by whether the speech had the ‘funny’ tag (1 = funny tag present ) or not (0 = funny tag not present) on the TED website. Due to humour has certainly an effect on how the content is perceived (Goodrich, Schiller, & Galletta, 2015). Humour, gender and topic were in a format that a correlation test could be conducted but age of the video needed to be recoded from the date format. The first month when they started to publish the videos got recoded as 0 then other videos got a number that was as higher as many months older they were.

Moderating variables

Regarding the variables of the moderators, variable ‘teaches something new’ (V19) and variable ‘it gives a new explanation or a new aspect of something’ (V20) gave the data on novelty. The levels of the three point scale were: 1 ‘not at all’, 2 ‘some new,’ 3 ‘to a great extent’ and it was compared to the knowledge of a university-educated, average person. On surprise, looking into surprising facts, results, examples in the text, (V21) and surprising acts (V22) were the two variables. Both were measured on a 3 points scale as well: ‘1, not at all 2, somewhat 3, a great deal.’ The variable novelty and surprise were also created. Novelty is the mean value of variable 19 (teaches something new) and variable 20 (it gives a new

explanation or a new aspect of something), while surprise got calculated from the mean of the two variables: surprise in the text and a surprising act. (Novelty M=2.12 SD=.57 Surprise M=1.47 SD=.54)

(17)

The dependent variable

The dependent variable was variable 23, the number of views, a proxy of virality. This information indicates how viral the video got, how many people it has reached. This

information is public on the website of TED.

Standardization

Because count data is usually not normally distributed (Field, 2013), all variables were standardized by creating a z-score. Logos is the least normally distributed with 5 cases, while ethos, pathos, novelty and surprise have hardly any cases outside the of the -1.96-1.96 range.

Results

In total 59 videos from the TED 2016 Conference were content analysed, their ethos, pathos, and logos, and on surprise and novelty content contained in the speech. Because this study is particularly interested in what content elements make a video go ‘viral’, causal relationships were examined between the content elements and the number of views.

Correlation analysis

A bivariate correlation analysis was run to check whether there is a relationship between the control variables (humour, gender, and video age) and the independent variables. Since the data is not linear, the Spearman rank-order correlation coefficient was calculated for our categorical data. The correlation test showed that only gender of the speaker

correlates statistically significant with the dependent variable number of views (r = .322, p = .013). Therefore, gender is included in all further analyses.

(18)

Regression analyses

Since the data is not linear, a negative binomial regression should be conducted to test for causal relationships between the variables (Field, 2013). In all analyses, the dependent variable is the number of views. The independent variables are ethos, pathos, and logos; the moderating variables are novelty and surprise; and the control variable in all the analyses is gender of the speaker.

Direct effect of logos, pathos, ethos on number of views.

The Omnibus test showed that our model was close to significant improvement over a null model (a model without predictors) but overall it was not significant with p = .092. To avoid overcorrection in the model, we set the value at the negative binomial regression with log from 1 what SPSS is using by default, to estimate value so we prevented the result to be overly conservative (Field, 2013). The dispersion parameter of the test became 0.405 as a result of the change of settings. A negative binomial regression with logos, pathos and ethos as independent variables and number of views as dependent variables showed that no significant main effect of logos b= .83 p = .695, nor a significant main effect of pathos, b= -.47, p = .576, nor a significant effect of ethos b = .13 p = .600 was found. It can be concluded that the three independent variables do not significantly predict the number of views.

Moderating effect of novelty and surprise

The negative binomial regression analyses with the interactions of pathos*novelty and then pathos*surprise as independent variables and number of views as dependent variable also did not find a statistically significant result. With former values being b=.52 p=.308 and b=-0.36 p=.466 for the letter one. Therefore, the moderators novelty and surprise do not have a statistically significant effect on the relationship between pathos and the number of views.

(19)

Control variables

Gender is the only variable that had to be controlled for in the analyses. Gender explains number of views b=.32 p=.033, which means if the speaker is male there is a 0.32 predicted increase for number of views. This also means that after including all of the

variables in a regression analysis, gender of the speaker remains the only significant variable.

Conclusion and discussion

This empirical research was the first attempt to investigate the relationship of public speeches and virality. This paper aimed to find whether Aristotle’s ancient means of

persuasion (ethos, pathos, and logos) drive speech public virality, as measured by the number of online views of TED talks. It hypothesised that pathos explains more variance in a

speech’s virality than ethos and logos. It was also expected that novelty and surprise will have an impact on this relationship and speeches with pathos and more novelty and/or

surprise are even more likely to become viral. To find these effects, a sample of 59 TED talks from the TED2016 Conference was content analysed and negative binomial regression analyses were performed. Specifically, the number of views was regressed on pathos, logos, ethos and the potential moderators novelty and surprise.

The findings showed that our hypotheses are not supported; pathos, ethos, and logos do not explain the number of views. This is also true for the moderator effect of novelty and surprise; it cannot be said that novelty or surprise would change the relationship to number of views of the speech videos due to the non-significant results. Therefore, the hypotheses must be rejected. However, gender is a variable that had to be controlled for did reveal a

statistically significant result, with an increase in gender causing the number of views to go up, if the speaker is male.

(20)

Theoretical implications

This study’s hypotheses were built on a line of argumentation from Aristotle. Many scholars proved that ethos, logos and pathos are persuasive (Brinton, 1988; Delia, 1970; Hass, 1981). This paper tested the means of persuasion in online environment and its relation to virality. From the results it can be concluded that Aritotle’s ideas in this particular context are not applicable. There is several evidence that his rhetorical theories work in the offline world (e.g. Halloran, 1982; McCroskey, 1966; McCroskey & Young, 1981). This paper expected that pathos has a more have a more significant influence on number of views. That emotions greatly affect human decision-making is well-known (Brinton, 1988; De Graaf, Sanders, & Hoeken, 2016; Leighton, 1985). Studies has also confirmed that positive emotions increase virality (Nahon & Hemsley, 2013). So a consequence, this study could tell three things. First, when speeches are in online environment the Aristotle’s means of persuasion are less relevant compared to an offline speech. Secondly, it could also be the case that Aristotle’s rhetorical guidelines work online, as well, in a sense that they are able to persuade one viewer but this somehow do not reflect in sharings and will not show a result on virality. Thirdly, this study did not focus only on positive emotions. Since positive and negative emotions were calculated together as pathos, this is maybe why this research cannot strengthen Nahon and Hemsley, (2013) findings that positive emotions in online content would lead to virality. Since this exploration was the first of its kind, the execution of the thesis is still relevant. There is certainly an area for further researches to investigate into different kind of contents what makes them go viral. Public speeches are performed in everyday life, as well; therefore it has a social relevance too, why should it be proven what works for an online audience. According to the argumentation, novelty and surprise do play a role in the likelihood of going viral (Eelen, Verlegh, & Van den Bergh, 2015; Teixeira, 2012;

(21)

Dobele et al., 2007) but regarding speeches this paper did not find a statistically significant proof of it. However, this study confirmed that gender plays a role in the number of views, male speakers do have better chances to go viral. The fact that it is easier to be persuasive for men has been proven already (Berryman & Wilcox, 1980). It is interesting to see that it is still true, according to this research and also gives men more views.

Practical implications

One of the possible practical reasons why the findings did not follow up the

argumentation and proved the hypotheses can be the operalization of variables. Capturing pathos, logos and ethos in a measurable form in a speech is not obvious, it has been done in different ways throughout the centuries of the science of rhetoric and it has not been in relation to virality at all, yet. It would be interesting to see - even though, the codebook was precisely defined and based on theories – that if we compute the pathos, logos, ethos

variables distinctively, would it still stand that Aristotle’s views are not applicable in the context of virality.

Limitations and Suggestions for Future Research

It is an interesting subject, combing an ancient science, rhetoric to a new field of communication, virality. Speeches do go viral, TED talks or commencement speeches for instance, the genre exist but what exactly makes them go viral that still need to be further proven in a future research. There are two suggestions for this, regarding the operalization of variables: first, more objective coding would further improve the exactness of pathos. Timing or exact numbers can give more accurate data than evaluation, even though, the coders were well-trained. Secondly, pathos, logos and ethos could be broken down even more. For

(22)

pathos. These, if using separately might give another result on their role in virility. Until the next research in the field, it can be concluded that Aristotle explains the attitude change in offline persuasion, but his classical ideas has little value in the online world of persuasion, i.e. virality.

References

Small, D. A., Loewenstein G., & Slovic P. (2006). Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims.

Organizational Behavior and Human Decision Processes, 102 (2007) 143–153 Hample, D. (1986). Logic, Conscious and Unconscious. The Western Journal of Speech

Communication, 50 (Winter 1986), 24-40

Aristotle (1946). Rhetorica. W. R, Roberts (Trans.). In Works of Aristotle (Vol 11) Oxford:Clarendon Press

Mangold, W.G., & D.J. Faulds. 2009. Social media: The new hybrid element of the promotion mix. Business Horizons, 52, 357–65.

Zeng, F., L. Huang, & W. Dou. 2009. Social factors in user perceptions and responses to advertising in online social networking communities. Journal of Interactive Advertising 10, no. 1: 1–13.

Van Noort, G., Antheunis, M. L., & van Reijmersdal, E. A. (2012). Social connections and the persuasiveness of viral campaigns in social network sites: Persuasive intent as the underlying mechanism. Journal of Marketing Communications, 18(1), 39- 53. Schwarz, N., & Clore, G. L. (1983) Mood, misattribution, and judgments of well-being:

Informative and directive functions of affective states, Journal of Personality and Social Psychology, Vol.45(3), pp.513-523

Hendriks Vettehen, P., Nuijten, K., & Peeters, A. 2008. Explaining effects of sensationalism on liking of television news stories. Communication Research, 35(3), 319–338.

(23)

De Graaf, A., Sanders, J., & Hoeken, H. (2016). Characteristics of narrative interventions and health effects: A review of the content, form, and context of narratives in health-related narrative persuasion research. Review of Communication Research, 4, 88-131. doi:10.12840/issn.2255-4165.2016.04.01.011

Cardon, M. S., Wincent J., Singh, J., & Drnovsek, M., (2009). The Nature and Experience of Entrepreneurial Passion. The Academy of Management Review, Vol. 34, No. 3 pp. 511-532

Bono, J.E., & Ilies, R., (2006). Charisma, positive emotions and mood contagion. The Leadership Quarterly 17 317–334

Brinton, A. (1988). Pathos and the “Appeal to Emotion”: An Aristotelian Analysis. History of Philosophy Quarterly, 5(3), 207–219.

Gallo, C. (2014).Talk like TED : the 9 public speaking secrets of the world's top minds London, Macmillan

April, A., Kedrowicz, J. & Taylor, L. (2016). Shifting Rhetorical Norms and Electronic Eloquence: TED Talks as Formal Presentations. Journal of Business and Technical Communication, Vol. 30(3) 352-377 DOI:

10.1177/1050651916636373

di Carlo D.S.,(2015). Stance in TED talks: Strategic use of subjective adjectives in online popularization, Ibérica 29 (2015): 201-222

Trevarthen, C. (1992). An infant’s motives for speaking and thinking in the culture in A.H. Wold (ed.), The Dialogical Alternative, 99-137. Oslo, Scandinavian University Press.

Sørensen, L. S. (2013). How to grow an apple: Did Steve Jobs speak Apple to success? - An analysis of Steve Jobs’ rhetorical and linguistic development in relation to Apple’s organizational performance. Masterthesis. Aalborg University.

(24)

Niebuhr, O., Voße, & J., Brem, A., (2016). What makes a charismatic speaker? A computer-based acoustic-prosodic analysis of Steve Jobs tone of voice. Computers in Human Behavior 64 (2016) 366e382

Hass, R.G., (1981) Effects of source characteristics on cognitive responses and persuasion. in R.E. Petty, T.M. Ostrom, T.C. Brock Cognitive responses in persuasion (pp 141-172). Hillsdale, NJ: Lawrence Erlbaum

Burgoon J.K., Kelley D.L. , Newton, D.A., & Keeley-Dyreson,M.P.(1989) The nature of arousal and nonverbal indices. Human Communication Research 16, 217-255. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L.

Berkowitz (Ed.). Advances in experimental social psychology (Vol. 19, pp. 123– 205). San Diego,CA: Academic Press.

Delia, J.G. ,(1970). The logic fallacy, cognitive theory, and the enthymeme: A search for the foundations of reasoned discourse. Quarterly Journal of Speech , Volume 56, 1970 - Issue 2

Hample D.,(1986). Logic, Conscious and Unconscious. The Western Journal of Speech Communication, 50, 24-40

Knight, C.M. (1999). Viral marketing – defy traditional methods for hyper growth. Broadwatch Magazine, Vol. 13 No. 1, pp. 50-53.

Leskovec, J., Adamic, L.A., & Huberman, B.A., (2007). The dynamics of viral marketing. ACM Trans Web, Vol. 1 No. 1, doi:10.1145/1232722.1232727

Mills, A.J., (2012). Virality in social media: the SPIN Framework. Journal of Public Affairs. Vol. 12 No. 2, pp. 162-169.

(25)

components of viral success. in Wuyts, S., Dekimpe, M.G., Gijsbrechts, E. and Pieters, R.(Eds), The Connected Customer, The Changing Nature of Consumer and Business Markets, Routledge, pp. 257-282.

Dobele, A., Lindgreen, A., Beverland, M., Vanhamme, J., & van Wijk, R. (2007). Why pass on viral messages? Because they connect emotionally, Business Horizons, Vol. 50, pp. 291-304.

Phelps, J.E., Lewis, R., Mobilo, L., Perry, D., & Raman, N. (2004), “Viral marketing or electronic word-of-mouth advertising: examining consumer responses and motivations to pass along email”, Journal of Advertising Research, December, pp. 333-348.

Wallsten, K. (2010), ‘Yes we can’: how online viewership, blog discussion, campaign

statements, and mainstream media coverage produced a viral video phenomenon, Journal of Information Technology, Vol. 7 Nos 2/3, pp. 163-181.

Arnould, E.J., & Thompson, C.J. (2005), Consumer culture theory (CCT): twenty years of research, Journal of Consumer Research, Vol. 31 No. 4, pp. 868-882.

Beverland, M., Dobele A., & Farrelly, F., (2015). The viral marketing metaphor explored through Vegemite. Marketing Intelligence & Planning, Vol. 33 Issue: 5, pp.656-674, doi: 10.1108/ MIP-08-2014-0146

Pitta, D. (2008). Providing the tools to build brand share of heart: gydget.com. Journal of Product & Brand Management, Vol. 17 No. 4, pp. 280-284.

Hirschman, E.C. (1980). Innovativeness, Novelty Seeking, and Consumer Creativity. Journal of Consumer Research, 7(3), 283–95.

Eelen,J., Verlegh, P. & Van den Bergh, B. (2015). Exploring the Effectiveness of the Label “New” in Product Packaging and Advertising. in NA - Advances in Consumer

(26)

Research Volume 43, eds. Kristin Diehl and Carolyn Yoon, Duluth, MN : Association for Consumer Research, Pages: 512-513.

Dobele, A., Lindgreen, A., Beverland, M., Vanhamme, J. & Van-Wijk, R. (2007). Why Pass on Viral Messages? Because They Connect Emotionally. Business Horizons, 50, 291-304. (DOI: 10.1016/j.bushor.2007.01.004).

Teixeira, T. (2012). The New Science of Viral Ads. Harvard Business Review. March, 2012, 25-27.

Meyer, W., Niepel, M. (1994). “Surprise,” Encyclopedia of Human Behavior, 4, 353–8. Derbaix, C., &. Pham, M. T. (1991),.Affective Reactions to Consumption Situations: A Pilot

Investigation, Journal of Economic Psychology, 12, 2, 325–55.

Rust, Roland T. & Oliver, R. L., (2000). Should We Delight the Customer? Journal of the Academy of Marketing Science, 28, 1, 86–94.

Vanhamme, J.,(2000). The Link Between Surprise and Satisfaction: An Exploratory Research on How Best to Measure Surprise. Journal of Marketing Management, 16, 6, 565–82.

Derbaix & Vanhamme, J., (2003). Inducing Word-of-Mouth by Eliciting Surprise—a Pilot Investigation, Journal of Economic Psychology, 24, 1, 99–116.

Berryman, C. L., & Wilcox, J. R. (1980). Attitudes toward male and female speech: Experiments on the effects of sex-typical language. Western Journal of Speech Communication, 44, 50– 59.

Cambridge Dictionary (2017)

Retrieved from: http://dictionary.cambridge.org/dictionary/english/fact

Schiller, G.K., Shu, Z., & Galletta,D. (2015). Consumer reactions to intrusiveness of online-video advertisements do length, informativeness, and humor help (or hinder)

(27)

marketing outcomes. Journal of advertising research, Vol.55(1), pp. 37-50 [Peer Reviewed Journal]

Bonchi F., Castillo C., & Ienco D., (2013). Meme ranking to maximize posts virality in microblogging platforms. Journal of Intelligent Information Systems 40:211–239. Tucker C., (2011). Virality, Network Effects and Advertising. Cambridge, MA: NET Institute,

MIT.

Urista M.A., Dong W., & Day K.D., (2009). Explaining why young adults use MySpace and but not on Facebook? How word-of-mouth through uses and gratifications theory. Human Communication 12(2): 215–229.

Phelps J, Lewis R, & Mobilio L, (2004) Viral marketing or electronic word-of-mouth

advertising: examining consumer responses and motivations to pass along email. Journal of Advertising Research 44(4): 333–348.

Corcoran, S. (2011, August 5.). Defining Earned, Owned and Paid Media. Sean Corcoran’s Blog Retrieved from:

http://blogs.forrester.com/interactive_marketing/2009/12/defining-earned-owned-and-paidmedia.html.

Akpinar, E., & J.Berger. (2017) Valuable Virality. Journal of Marketing Research Vol. 54 Issue 2, p318-330.

Alhabash, S., & McAlister A.,R., (2015). Redefining virality in less broad strokes: Predicting viral behavioral intentions from motivations and uses of Facebook and Twitter. New Media and Society2015, Vol. 17(8) 1317–1339

Yuki, T. (2015). What makes brands’social content shareableon Facebook?. Journal of Advertising Research, 55(4), 458-470.

Eisingerich, A. B., Chun, H. H., Liu, Y., Jia, H. M., & Bell, S. J. (2015). Why recommend a brand face-to-face but not on Facebook? How word-of-mouth on online social

(28)

sites differs from traditional word-of-mouth. Journal of Consumer Psychology, 25(1), 120-128.

Petek, T.(2012). Criteria for public speech planning : characteristics of language learning. Linguistica, 01 December 2012, Vol.52(1), pp.381-392

Lagerwerf, L., Amber Boeynaems, A., Egmond-Brussee, C., & Burgers C.(2015). Immediate Attention for Public Speech Differential Effects of Rhetorical Schemes and Valence Framing in Political Radio Speeches.Journal of Language and Social Psychology Vol 34, Issue 3, 2015

Holtgraves T., & Lasky B. (1999). Linguistic power and persuasion. Journal of Language and Social Psychology, 18, 196-205.

McCroskey, J.C., (1966). Scales for the measurement of ethos. Speech Monographs Volume 33, 1966 - Issue 1

McCroskey, J. C., & Young, T.J.(1981). Ethos and credibility: The construct and its measurement after three decades. Central States Speech Journal Volume 32, 1981 - Issue 1

Halloran, S. M. (1982) Aristotle's Concept of Ethos, or If Not His Somebody Else's Rhetoric Review, Vol. 1, No. 1,pp. 58-63

Lucas, S. E. (1971). Notes on Aristotle's Concept of Logos. Quarterly Journal of Speech. Anton, J. P. (1997). Aristotle on the Nature of Logos The Society for Ancient Greek

Philosophy Newsletter. 243. Newsletter 3-1997

Rees, B. R. (1972). Pathos in the Poetics of Aristotle. Greece & Rome Volume 19, Issue 1, pp. 1-11

(29)

Schamari, J., & Schaefers, T. (2015) Leaving the home turf how brands can use webcare on consumer-generated platforms to increase positive consumer engagement Journal of interactive marketing : a quarterly publication from the Direct Marketing Educational Foundation, 2015, pp. 20-33 [Peer Reviewed Journal]

Karine, N., & Hemsley, J. (2013) Going Viral Cambridge, Polity.

Leighton, S.R. (1985). A New View of Emotion. American Philosophical Quarterly, vol. 22, pp. 133-14.

Burt, R. S. (1999). The Social Capital of Opinion Leaders. The ANNALS of the American Academy of Political and Social Science, 566(1), 37–54.

https://doi.org/10.1177/000271629956600104

Krippendorff, K. (2004). Content Analysis: An Introduction to Its Methodology. London, Sage Publications

Gallo, C. (2014), Talk like TED: 9 public speaking secrets of the word’s top minds. New York, St. Martin’s Press

Field A., (2013) Discovering Statistics Using IBM SPSS Statistics. University of Sussex, SAGE Publications Ltd

Appendix

Codebook

Administrative data v1 Speech ID

Identification number of the video as assigned in the Excel Sheet. _ _

(30)

1 Timea 2 Karishma 3 Balazs

General information on the speech v3 Gender of the speaker

0- female 1- male 2- other

v4 Is she/he a famous person (e.g. celebrity)?

Is the person globally well-known also beyond his profession? examples for famous person: Serena Williams, Bill Gates

0-no 1-yes

v5 Length of the speech (example: 12:45)

v6 When was the video published online? Info available under the video before you open it POSTED not Filmed!

example: March 2015 (month year)

(31)

v7 Topic of the speech

Which category does it fit the most? Only pick the most relevant one! 1-Technology 2-Entertainment 3-Business 4-Science 5- Environment 6-Minority rights

7- International relations, state 8- Motivation

9- Art 10- Lifestyle 11-Other

v24 Is the video humorous?

Is the funny tag there under the video on the TED website? 0: no

(32)

Role of modes of persuasion in the speech Logos:

v8 Reasoning: The speaker used logically built arguments the thought is an argument , if it has these three components: 1.premises (=statements)

2. method of reasoning (e.g. inductive: from general principle to particular case, deductive: from particular case (observation) to general principle and abduction: incomplete observation with the best possible explanation)

3. conclusion

Argument examples:

not really: 0-1

moderately used: if a speech is 6-10 minutes long: 2 if a speech is 10-15 minutes long: 2-3 if a speech is 15-21 minutes long:2-4

used a lot: if a speech is 6-10 minutes long: 3 or more

(33)

if a speech is 15-21 minutes long: 5 or more

Note: if the argument is really complex and long then it counts as +1 (so if there was 2 argument in a 7 minutes long speech and one of them was explained over 2 minutes with several premises etc. then it is worth 3 arguments in total and goes to the “used a lot” category.

1– not really

2- moderately used 3-used a lot

v9 The speaker cited scientific facts in his speech

a scientific fact is any observation that has been repeatedly confirmed and accepted as true; any scientific observation that has not been refuted (Source: Dictionary.com's 21st Century Lexicon)

A scientific statement that stands without explaining the whole argument.

Example: The earth is round.

not really: 0-1

(34)

if a speech is 10-15 minutes long: 4 or more if a speech is 15-21 minutes long: 5 or more

1– not really

v10 The speaker used statistics in his speech

The speaker used numbers (visually or verbally) that represent information. not really: 0-1

Note: if the explanation of a statistical figure is really complex and long then it counts as +1 (so if there was 2 stats data in a 7 minutes long speech, and one of them was explained for a long time then it is worth 3 stats in total and goes to the “used a lot” category.

(35)

v11 The speaker used literary and/or historical analogies in his speech

The speaker –as an explanation or example- compared something to a literary and/or historical event, process, fact.

not really: 0-1

Note: if the analogy is really long then it counts as +1 (so if there was 2 analogies in a 7 minutes long speech, and one of them was explained for a long time then it is worth 3 analogies in total and goes to the “used a lot” category.

1– not really

(36)

Pathos:

v12 The speaker used stories in his/her speech.

story or narrative: at least one character experiences at least one event. (Character doesn’t need to be a human, can be a state or a group etc.)

not really: 0-1

Note: if the story is long then it counts as +1 (so if there was 2 analogies in a 7 minutes long speech, and one of them was long then it is worth 3 stories in total, if both of them were long then 4 stories in total. )

1– not really 2-moderately used 3-used a lot

(37)

a metaphor describes a person or object by referring to something that is considered to have

similarcharacteristics to that person or object. examples: "The mind is an ocean" and "the city is a jungle" (Source: Cambridge Dictionary)

not really: 0-1

v14 The speaker appealed to negative emotions

when the speaker recognizably aims to evoke negative emotions (e.g. to fear, anger, guilt, disgust) in the audience.

Format: X must be true. Imagine how sad it would be if it weren’t true. Example: There must be objective rights and wrongs in the universe. If not, how can you possibly say that

torturing babies for fun could ever be right? (Source: www.logicallyfallacious.com)

(38)

v15 The speaker appealed to positive emotions

when the speaker recognizably aims to evoke positive emotions (desire, pride, hope) in the audience for supporting his/her claim.

Format: X must be true. Imagine how good it would be if it was true.

not really: 0-1

(39)

Ethos:

v16 The speaker talks about his professional background (directly or indirectly) only count those things when he mentions stuff what are relevant for why he is giving this speech/why he is an expert on the topic he is speaking about, and not the just personal stories from his life.

not really: 0-1

(40)

The speaker seemed enthusiastic about his/her speech/topic (for example: he/she had a fast tempo that varied a lot and/or higher pitches)

1– not really 2-moderately 3-a lot

v18 The speaker looked confident

His body language and facial expression was supporting, he did not show sign of nervousness. for example: touching the head

1– not really 2-moderately 3-a lot

Factors of virality Novelty

v19 The speech teaches something new

(compared to the knowledge of an university-educated, average person) 1 not at all

2 some new 3 to a great extent

v20 it gives a new explanation or a new aspect on something a common topic is explained from a new perspective

1 not at all 2 some new 3 to a great extent

(41)

Surprise factor

v21 Was there anything shocking/surprising in the speech? in the text of the speech (e.g. facts , results, turn in the speech) 1 not at all

2 somewhat 3 a great deal

v22 Was there any shocking/surprising act during the performance, which was unexpected?

for example: Bill Gates releases mosquitoes into audience in his TED talk 1 not at all

2 somewhat 3 a great deal

Number of views v23 How many times was the video viewed?

you can find this info under the video (No:_ _ _ _ )