The history of journalism research in just 5 networks

Capture d’écran 2016-01-25 à 10.11.49I know this title is very pretentious but I just wanted you toclic on the link. And it worked didn’t it? My bad… I’m just part of the “double clic” [1] generation. It seems that Mapping you field of research is another thing to do for any young researcher interested in DataViz. Building a network and providing data on, for example, important researchers in your disciplines, the most published people or papers or on the popular topics people are working on has become a requirement if you want to be part of the cool kids. This guy did it wonderfully with famous philosophers. Martin Grandjean used Twitter data to detect researchers and professionals involved in DigitalHumanities. The truth is that it is a good exercise for two main reasons: (1) if you are a researcher you must know a little about your field  and what it looks like and (2) academic data is usually more available than data on mafia mobs.

So here is my shot at it, The history of journalism research in keywords [2].  What I did was trying to find and represent the most used and connected keywords within journalism studies. The results is five graphs, representing a different period of time. To do so, I used Scopus as datasourcethis tutorial, as well as this video and these websites (1 & 2), and Gephi of course to visualize it all. Please keep in mind that this data is incomplete because coming from a single directory (Scopus).


Before showing you the graphs, let me summarize the process. I first went on Scopus, which is a database of scientific literature and pulled all the results for the keyword “journalism”. That gave me 12398 published scientific papers. In order to be able to see the evolution of the keywords, I intuitively sampled the results to get groups of around 1000 articles. The four groups are (1) articles from 1977 to 1994, (2) 1996 to 2000, (3) 2005 and 2006, (4) 2010 and (5) 2015. Here is some additional data:

  • (1) from 1977 to 1994: 693 ARTICLES / 98 NODES / 225 LINKS.
  • (2) from 1996 to2000: 897 ARTICLES / 56 NODES / 112 LINKS.
  • (3) for 2005 and 2006: 967 ARTICLES / 402 NODES / 1495 LINKS.
  • (4) for 2010: 976 ARTICLES / 237 NODES / 786 LINKS.
  • (5) for 2015: 881 ARTICLES /  404 NODES / 1835 LINKS.

After this, I downloaded the 5 databases  including the keywords given by the authors (not the keywords from Scopus) and in the CSV format. I then used the ScienceScape tool from SciencesPoMediaLab to create networks of keyword co-appearing in the same paper. A link/edge is created whenever the keyword appears with another keyword in the same paper. The node size depends on the number of occurrences (the bigger the node, the most it was used by researchers). The colors of the nodes were established using the modularity algorithm and show the communities of keywords (keywords often cited together). For the first graph, I kept all  keywords (because they weren’t many) and only kept those with two occurrences or more for the four following graphs. This is the reason why the first graph has more keywords than the second (in reality, it doesn’t). This has nothing to do with journalism, just with the fact that researchers didn’t use keywords as much before or that they were not always indexed. Keeping only the keywords with two occurrences or more would have given me a first graph with 5 o 6 nodes. Additionally, I used ForceAtlas as spatialization (with the linglog mode) and took the keywords “journalism”, “Media” and “News” out of the networks for obvious reasons (FIY: they were used a lot).



Of course, I must admit that the results are partial and biased. The changes in the number of keywords I kept (according to occurrences) and the weird sampling might make some of you uncomfortable. Plus Scopus is just a database and does not have all articles by all publications (though the results I got come from a variety of well known publications such as Journalism Studies, the Journal of Communication, etc.).

Nonetheless, I found the results very interesting. They indicate that Between 1977 and 1994 researchers focused on black press issues (probably especially in the US) and on other themes related to reporting on war, violence, and terrorism such as in China. You can see that by looking at the really light cluster of nodes on top of the graph and the light orange cluster in the centre. Research also focused on media and mass media specifically the rôles of said media (as a watchdog for example).

Things seem to start changing between 1996 and 2000 where the keywords used relate less to a theme or a topic covered by the media and more on the different types of media (mass media, news media, television), their values (objectivity, civic engagement) and links to topics that come from other discipline (or so it could be argued) such as political communication, public sphere, public policy, popular culture, gender, etc. We also see the apparition of something that will influence keywords for years to come: the word “internet”.

In 2005 and 2006, we can observe a big keyword boom (which does not mean a big research boom, let me insist on this). The usual topics are stil represented (television, democracy, news papers, objectivity, mass media, etc.) but I see a boom in keywords that link Journalism and the Internet (online journalism, web 2.0, online newspapers, convergence, citizen journalism, blogs). Also interesting is the apparition of the word journalistS with an “S”, as well as professionalism, gender, etc.  This is naturally my own interpretation but I interpret this as shift towards more interdisciplinarity and towards a more human approach. It sure looks like journalism scholars have stopped trying to integrate hard sciences and asked to join the social sciences… or something of the sort, this needs to be developped. Journalism is not just this content we get to analyze to interpret public events anymore, journalism is also a profession, a social practice that involves peoples, it is a category of the social world that we need to reflect upon.

2010 looks a lot like 2005 and 2006 with small differences. This make sense since these years aren’t to far appart. However, I do see a comeback of topics related to conflict reporting (word like conflict reporting, war reporting, foreign correspondent) and national coverage (keywords such as Brazil, Israel, Africa, South Africa). This also indicates that more and more research in these big journals is dedicated to countries outside of the US and western Europe. Online journalism and research related to the Internet as now taken a big place, occupying the center right of the graph and upper right corner. The keywords with “online” in it (online journalism, online news, online newspapers, online media) and their clustering in a community tend to indicate that a sub-discipline of the field that focuses on online media and news might be emmerging.

In 2015, keywords tend to be  widely interconnected, which could indicate an hybridization of research themes, topics and affinities; though I have no way to explain how and why this is the case. This could again just be a change in keywording practices by researchers. Having said this, it seems that topics related to the Web occupy almost half of the graph.  The keyword “internet” however has reduced in size which shows how we are deeper into the matter now. This is also shown by the smaller size of the node online journalism and the apparition of the keyword digital journalism. Finally, social media has, of course, taken a central place in research.  Interestingly enough, Twitter has a much bigger place in this area than Facebook. Could this be a bias of researchers because of data accessibility ? Or is this just because journalists and audiences use twitter more than Facebook and other social networking sites… I have my opinion but I’ll let you be the judge of that.

So… What do you think? Would you have interpreted these graphs the same way? Feedback is always appreciated, and do not hesitate to leave a comment to get  full data and graphs.


[1] This is for all my (3?) Latourian friends out there ![2] By the way, the data collected and the analysis hereafter have do not hold up to scientific standards nor did I want them to. It is simply an exercise I did for myself but that I now want to share.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s