Notes

While there is enormous diversity in communication style within each gender group, in this post I want to focus on an interesting phenomenon that emerges in Political Communication between genders and in particular on rhetorical devices connected to verbal "aggressiveness".

For the analysis we considered a corpus of roughly 3.600 speeches. These are transcripts of political speeches tagged with audience reactions, such as APPLAUSE or LAUGHTER. In our model, different audience reactions can reveal different persuasive strategies used by the speaker. The "Negative-Focus" tag group is of specific interest here; it includes phenomena such as BOOING. This group of audience reactions indicates a persuasive attempt that sets a negative focus in the audience, where the negative focus is set towards the object of the speech and not on the speaker themselves (e.g. S: “Do we want more taxes?” A: "No!"). This group of tags is interesting since it signal a more "aggressive" kind of rhetoric, where the speaker brings the audience to explicitly demonstrate their negative feelings. For the analysis we use the concept of "Tag-Density", a measure that indicates how often audience reaction tags appear in the speeches under scrutiny.

And here are the unequivocal results: while all other tag densities are almost the same, for the negative-focus tags we have a density 60 times higher for male speakers.

Interestingly, similar results are found for the Democrats/Conservatives partition, with the rhetoric of the former being much less "aggressive" than the rhetoric of the latter.  

Further details in: Guerini, M., Giampiccolo, D., Moretti, G., Sprugnoli, R., & Strapparava, C. (2013). The New Release of CORPS: A Corpus of Political Speeches Annotated with Audience Reactions. In Multimodal Communication in Political Speech. Shaping Minds and Social Action. Poggi, I., D’Errico, F., Vincze, L., & Vinciarelli, A. (eds.).  pp. 86-98. Springer Berlin Heidelberg.

What is the optimal post length – either on Twitter or Facebook – for increasing engagement on these Social Networks? A couple of white papers from TrackSocial recently found that on Facebook and Twitter the ideal length is around 70-100 char. What strike me most is that, apart from minor differences, the ideal length does not vary across social platforms: 

1) On Facebook the optimal length is around maximum tweet length (after 140 char engagement drops drammatically)

2) On Twitter the engagement is around 70-100 chars, quickly decreasing off these values. 

The possible explanation is that, given users' "rapid cognition" approach to the overwhelming information stream, succint posts are more likely to convey their message before users' attention drops. This model also explain why including photos drastically increase post virality (photos grab the attention with little or no effort at all). 

In the world of Viral Marketing – as well as in the social networks analysis field – there is the tendency to assume that social contagion is a uniform phenomenon, whose dynamics are always the same for any content in any context (at least at the macro level). But are we sure that it is really enough to create a viral content and to deliver it to the adequate opinion leaders for obtaining the desired effect, without worrying about the micro level? A series of experiments seems to refute this view, suggesting that a more articulated approach is needed. According to our findings, traditional approaches based on “popularity metrics” (e.g. the number of I like) are not sufficient. Virality is a multi-faceted phenomenon, which can bring about different audience reactions depending on the characteristics of the delivered content (exactly in the same way different viruses can provoke different symptoms). Reactions, of course, must be carefully considered in planning a successful promotion campaign. Let's see in detail what are these different forms of "viral symptoms":

  • Appreciation: how much people like a given content, for example by clicking an "I like" button.
  • Spreading: how much people tend to share this content by forwarding it to other people.
  • Simple buzz: how much people tend to comment a given content.
  • White buzz: how much people tend to comment in a positive mood (e.g. “The best product I have ever bought”).
  • Black buzz: how much people tend to comment in a negative mood (e.g. “Do not buy this product, it is a rip-off”).
  • Raising discussion: the ability to induce discussion among users, rather than on the content itself.
  • Controversiality: the ability to split the audience in different parties (usually pro and against the given content).
  • Fostering elaboration: the ability to induce the audience to elaborate on the given content, by writing long comments.

To give ground to our claims, we provide initial experiments in a machine learning framework to show how various aspects of virality can be independently predicted according to content features. A psyco-linguistic analisys further supported this view showing that different linguistic styles bear different effects. Further details can be found in: 

Guerini, M., Strapparava, C., Ozbal, G.: Exploring text virality in social networks. In: Proceedings of ICWSM’11. 2011. (PDF)

Strapparava, C., Guerini, M., Ozbal, G.: Persuasive language and virality in social networks. In: Proceedings of ACII ’11. 2011.

Why are certain pieces of online content more viral than others? And in particular, which role emotions play in this context? Jonah Berger and Katherine L. Milkman takes a psychological approach to understanding diffusion. Focusing on a dataset of the New York Times articles published over a three month period, the authors examine how emotion shapes virality (the metric used to measure virality is the number of people who emailed the article).

Findings are clear: positive content is more viral than negative content, but the relationship between emotion and social transmission is more complex than valencealone, and is partly driven by arousal. Content that evokes either positive (awe) or negative (anger or anxiety) emotions characterized by high arousal is more viral. Content that evokes low arousal emotion (sadness) is less viral. 

So use emotions if you want your content to go viral – but avoid sadness!

References: Berger, J. A., and Milkman, K. L. Social Transmission, Emotion, and the Virality of Online Content. Social Science Research Network Working Paper Series. 2009.

MarcoGuerini.eu

PhD in Information and Communication Technologies. Personal website.

©2018 Marco Guerini. All Rights Reserved. Web designed by GeekSolution

Search