Publication type: Conference other
Type of review: Peer review (abstract)
Title: Capturing media resonance through corpus-based discourse analysis : method development using the example of energy and social assistance discourse
Authors: Krasselt, Julia
Klopfenstein Frei, Nadine
Saner, Mirco
Bubenhofer, Noah
Calleri, Selena
Rosenberger Staub, Nicole
Wyss, Vinzenz
et. al: Yes
Conference details: Digikomm 2019, Berlin, 6.-8. November 2019
Issue Date: 7-Nov-2019
Language: German
Subjects: Media resonance; Corpus-based discourse analysis; Multi-system relevance; Digital linguistics; Journalism; Organizational communication
Subject (DDC): 070: News media, journalism and publishing
Abstract: Media releases are both: a PR tool for creating media resonance and an important source of information for journalists. The journalists, on the other hand, draw the attention to the statements and demands of the sources (Fengler/Russ-Mohl 2014: 249). Thus the PR system and the journalistic system are in an «interefficational relationship», where inductions and adaptation processes are taking place simultaneously on both sides, which in turn influence each other (Bentele/Nothaft 2008: 36; Bentele 2014: 357). Inductions are intended, directed communication stimuli or influences that emanate from one side and can lead to observable effects on the other. Both sides can only influence the other side by adapting to its structures, routines, knowledge and expectations (vgl. Schweiger 2013: 97). Through context control PR tries to work on conditions that can trigger favourable developments for the organisation in the environment (Nothaft / Wehmeier 2009: 162f.). To this end, the organisation must be present in the public discourse, refer to other contributions to the discourse and achieve resonance for its own contributions in the public space. For the position of the organisation in the public discourse it is essential whether the organisation can be regarded as legitimate or not by its stakeholders (Sandhu 2014: 1172). According to systems theory, journalism takes up topics and events that show mutual irritating references between different system rationalities (Kohring 2004: 188). If a communicated construction of reality is relevant for more than one system, it generates more need for discussion and therefore it is more likely to be taken up by journalism (Wyss 2011: 36; 38; Kohring 2006). In contrast to classical input-output analyses, the authors have developed a method using automated discourse analysis to investigate the mass media resonance of media releases on a textual level (citations) and on a thematic level (multi-system relevance). This method has been used in two interdisciplinary research projects in the context of the energy and the social assistance discourse in Switzerland. The automatic identification of media resonance is done in two parts: (1) extracting the parts of a PR statement which have been reused, cited or commented on in an article, (2) identifying the theme in the PR statements and the articles. In order to achieve (1), an algorithm has been developed to dynamically find the longest common sequences in multiple texts. The method is based on precomputing reoccurring sequences and their longest combination, and then finding duplicates in different texts (Ahonen-Myka 1999a,b). (2) has been implemented by automatically annotating the systems in a text. For each word the system looks up the assigned system(s) in a coded list. The list is based on the Dornseiff-Wortschatz (Dornseiff 2014), which was then manually coded and enriched for the specific discourses. Each word was assigned to at least one system. Owing to the ambiguous and contextual nature of meaning, the algorithm does not only look up the word itself but also words with a semantically similar profile in order to make the annotation less error prone. The semantically similar words are computed based on a word embedding model (WEM) (Mikolov 2013). A WEM consists of a vector-representation of each word in a corpus based on its contextual usage. Words used in a similar context have a similar vector and are thus in a similar semantic field. This can be utilized to solve two problems, first: not every word in a text is in the coded list, yet similar words could be and second: by checking the semantic fields of each word in a text, polysemous words can be disambiguated if the most similar words are categorized in only one system and the base word was categorized to reference multiple systems. The distributions of the referenced systems in each text are then computed in a bottom-up fashion, from sentence level to text level. This results in the annotation of reused sequences and the relevant system(s) in a text, which allows for a systematic analysis of input-output in combination with the multi-system relevance.
Fulltext version: Published version
License (according to publishing contract): Not specified
Departement: Applied Linguistics
Organisational Unit: Institute of Applied Media Studies (IAM)
Institute of Language Competence (ILC)
Published as part of the ZHAW project: RaPEnDi - Radar Public Energy Discourse
Appears in collections:Publikationen Angewandte Linguistik

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.