

405
Saturday, November 12
0 9 : 0 0 – 1 0 : 3 0
PP 531
Diversity in the Norwegian News Media Landscape: A Semi-Automatic Big Data Analysis of Diversity Indicators in the News
T. Pedersen
1
, H. Sjøvaag
1
1
University of Bergen, Information Science and Media Studies, Bergen, Norway
This paper presents findings from a semi-automatic big data content analysis of 189 news outlets in Norway, where the aim is to ascertain the extent
to which the national news landscape can be characterized as diverse. Diversity is operationalized as frequency and distribution in voices and topics in
the news, analyzed according to three structural factors: ownership affiliation (corporate or independent), medium type (legacy newspaper, pureplayer
or broadcasting), and distribution level (local, regional or national). The analysis finds that diversity in topics is primarily contributed by public service
broadcasting and the legacy regional and specialist press, and that diversity of voices is primarily supported by local newspapers.The analysis also finds that
corporate ownership and organizational resources matter for newsrooms’contributions to overall diversity. Diversity is one of the main measures to inves‑
tigate the internal and external diversity levels in media systems. As such, diversity is one of the aims of media policy. Content analysis of voices and topics
in the news enables evaluations of the extent to which representation of political opinions and identities is present in the editorial discourse. In this study,
natural language processing algorithms are used to extract named entities (persons and organizations) in news stories, while automatic content analysis
is mobilized to analyze content categories in the news. The data was collected using custom scrapers gathering 442 GB of html files from 189 individual
news websites from across the country between 1 October and 31 December 2015. Manifest variables (date, time, author, word count, links etc.) and latent
variables (content categories) were extracted using custom written algorithms. Evaluations of diversity levels in the Norwegian news media landscape
are operationalized according to the principle of ‘most different’. Venturing from the assumption that mediated agendas tend to move towards a common
conversation, the analysis aims to uncover the parameters that account for the presence of diverse voices and topics in the news. The ‘most different’voices
in this case lie outside the range of the political and organizational mainstream, representing women, minorities, small political parties, non-capital voices
and ‘ordinary’people, differentiated using named entity recognition. The ‘most different’topics entail locally affiliated stories and stories outside the main‑
stream news agenda, differentiated by proximity measures of geographical centre-periphery dimensions and natural language processing algorithms
identifying topical overlap illuminating agenda setting patterns in the national sphere.The analysis present a comprehensive picture of the voices and topics
represented in Norwegian news media in the period as a distribution of a selection of salient actors/topics across the various media outlets. Results show
that state supported editorial organizations, local newspapers and strong regional newspapers contribute most to the diversity of the Norwegian mediated
public sphere, and that corporate ownership represents a homogenizing factor. As findings indicate that external pluralism matters for internal pluralism,
the analysis should inform future media policy in Norway.
PP 532
Comparing Television News Sensationalism in 15 European Countries: The Importance of Media System Level Characteristics
K. De Swert
1
, I. Kuypers
2
1
University of Amsterdam, Communication Science, AMSTERDAM, Netherlands
2
University of Antwerp, Political Science, Antwerp, Belgium
Contrary to the expectations of prophets of doom and gloom, public and commercial television broadcasters are holding their key position in information
provision to citizens very well in times of technological changes, globalization and commercialization. More convincing (and worrying) prophecies focused
on the declining quality of television news.To keep attracting viewers, quality of television news would have to go down (e.g. McManus, 1994).The concept
of sensationalism was coined in this context and studied by various scholars in different contexts (see Kleemans & Hendriks Vettehen (2009) for an over‑
view). While studying various content elements of the news, however, the conclusions often are that there are few signs of such a general decline. This
leaves the field with many unanswered questions, not in the least about the importance of media system characteristics (like commercialism and audience
fragmentation) that are main suspects to stimulate the development of sensationalism. To account for those, there is a pressing need for more international
comparative research in communication science (Slater, Snyder, & Hayes, 2006) and in particular in sensationalism research (Hendriks Vettehen, Zhou,
Kleemans, D’Haenens, & Lin, 2012). And, the studies that do adapt such an approach, often work with relatively old data and suboptimal country choices
for this purpose (e.g. Cohen, 2013) or a very limited set of systems (Kleemans, Van Cauwenberghe, D’Haenens, & Hendriks Vettehen, 2008). This study will
try to fill part of that gap. Based on a fresh, large-scale content analysis (21 days in a constructed sample in the February-May period of 2016) of flagship
television news broadcasts of 30 television stations (the most watched public and commercial news broadcast per country) in 15 European countries, this
study will apply a multi-level approach to try to answer the following research questions: 1) What is the impact of media system level characteristics as
dependency on commercial revenues and audience fragmentation on the level of sensationalism? 2) What is the impact of public broadcasting on the level
of sensationalism? In order to answer these questions, sensationalism will be defined, divided up and measured following Vettehen, Nuijten, & Beentjes
(2005) as topic content sensationalism, formal features sensationalism and vivid storytelling sensationalism. Each of these forms of sensationalism will be
investigated separately, since studies in the past have revealed important differences in development between them. Since data collection for this project is
still ongoing, there are no results to be reported yet at this point. These will certainly be available (and complete) by the time of the conference. Inspired by
earlier work of e.g. Chan & Lee (2013) about the impact of public and commercial channels, and Arbaoui & colleagues (forthcoming) on media system char‑
acteristics, we speculate to find that mainly audience fragmentation is a driving factor for topic and format sensationalism (and less for vivid storytelling),
as well as we expect public channels to show less sensationalism than commercial channels, stressing the importance of public service news broadcasters
in an age of high competition and audience fragmentation.