Exploring Semantic Networks

Networks in times of social crisis: The Coca-Cosa Case in Colombia 

The sociologist Ronald S. Burt (1975) outlined a methodology for studying a society in transition as reflected in its mass media. He was concerned with the transition of the United States from a person oriented society in the 19th century to a corporate society during the 20th century and his strategy was to identify the networks of associations among categories of actors by using past issues of The New York Times. In this blog post I'm exploring how a variant of Burt’s method might be applied to study the transition of a society in times of social crisis using a combination of words parsing and social network analysis.

Under the assumption that these changes are reflected in the mass media (Alexander, 2006) and in order to collect relational data of the categories of actors interacting in the social structure across time, I have examined archival records from the media which reflect the transition of social structure in times of crisis. The society analyzed is Colombia and the mass media records examined is a corpus of 43 articles about the case of Coca-Cola published in El Tiempo (one of the most popular newspapers in Colombia) between 2002 and 2009. By this time, The Coca-Cola Company and its local bottlers in Colombia were accused of being associated with illegal paramilitary groups to threaten and kill some labor unionists in this country. 

My first task was to identify the relevant actors in the Coca-Cola case. After reading the articles I found that most of them where corporate actors such as The Coca-Cola company and its representatives. However, individual entities were also in these texts as in the case of some labor unionists. I used the words’ parser available at the software Atlas.ti 7.1 to identify the agentic terms as well as the frequency and co-ocurrence of such terms within each text. By using it, I found 73 agentic terms with an occurrence of 550 times between 2002 and 2009 excepting 2008 when any article was recorded in the national mass media. In this blog post will refer to these agentic terms as actors.

Figure 1 shows that corporate actors and latent corporate actors have received the most of the attention in the news with an average of 40% in the case of corporate actors and 37% for latent corporate actors. Beginning with 57% in 2002, corporate actors have received considerable attention, particularly in 2004 with about 92% of the news. In this category are The Coca-Cola Company and its representatives as Donald Knauss, Kari Bjorhus and Kerry Kerr. However, in 2006 latent corporate actors received a higher attention with about 39% of the news. Here is the participation of university students, labor unionists, human rights activists and also paramilitary groups. Both categories continued receiving the most of the attention. At a lower level, agents and persons have received an average of 22% and 1% of the attention respectively. But while persons were absent during the most of the time, agents received an important attention in 2005 and 2006, acting as carrier groups or means of interaction between the other categories of actors. 

Now, instead of looking at separate network characteristics, it may be more insightful to look at differences over the years. As figure 2 shows, the rise of corporate actors and latent corporate actors in 2005 and 2006 changed considerable the social structure as reflected in the media.

The graphs suggest that the structure of the network becomes fragmented when the participation of corporate actors and latent corporate actors increases and when they start to move from the periphery to the core of the network.  Later, when they move back to the periphery, networks return to be closed and dense again, while relations become stronger as denoted in the strength of the ties in 2009. 

With this, it seems that the participation of certain actors affects the structure of the network. To explore this, I performed a correlation analysis using Stata 11.2 in which I found that the type of actors and the structural indicators are related. The participation of corporate actors is positively related with betweenness (r= 0.77, p˂.005) and degree (r= 0.94, p˂.005) but negatively related with density (r= - 0.94, p˂.005) and closeness (r= - 0.78, p˂.005). Similar relations were founded between latent corporate actors and degree (r= 0.97, p˂.005), density (r= - 0.78, p˂.005) and closeness (r= - 0.85, p˂.005). The participation of agents is also positively related with degree (r= 0.80, p˂.005) and negatively related with density (r= - 0.96, p˂.005).

At this point, I will focus the attention on two major results: first, corporate actors and latent corporate actors are key players in times of social crisis. Second, they extend crisis by creating a fragmented society with a core/periphery structure in which they become betweenners or intermediary actors who control the resources and manage the information by laying in the paths of the other actors. Density and closeness seem to rise after the crisis has emerged and a cultural consensus about such a crisis as threatening collective identity is obtained.