A Corpus-Based Study on the Rhetoric of Persuasion in Science Blogs and Online Science News

Scientific advances do not make the news anymore, yet we rely on the news to keep up-to-date with everything that goes on in those areas of knowledge we are curious about. Since the Internet became a commodity, we have gained an incredible amount of science popularization websites, whose authors range from research institutions to passionate laymen, and it has never been so easy to feel experts in anything without years of training. All this availability of information, spread in particular by science news and scientific weblogs, has, however, at least one catch: identifying reliable sources. We take for granted that science and science-related news are inherently objective, and that any new piece of information we get is factually true. Yet both science and the news tend towards subjectivity as they both are bound to market constraints for funding (for example: private companies sponsoring and promoting research in their interests, institutions advertising their work through news campaigns). The aim of the present research is to find out if online scientific reporting is as subjective as science blogging when it comes to persuading readers to take a specific viewpoint on the topics they accessed. More specifically, the research investigates if the language used in news and blogs can be taken as proof that both journalists and bloggers convey opinion and facts at the same time. To this purpose, I decided to build two comparable corpora from the web (Gatto 2014; Schäfer & Bildhauer 2013): a Science News Corpus (SciNews) and a Science Blogs Corpus (SciBlogs). The corpora were compiled relying on two aggregators to select a sample from the overwhelming quantity of data available from the web, namely (e)ScienceNews (esciencenews.com) for science-related articles and ScienceSeeker (scienceseeker.org) for science-related blogs. Considering websites from any English-speaking country, in order to obtain a representative sample of full texts of news articles and blog posts I have further restricted the sampling in the following ways. As for SciNews, I decided to leave out all those news websites that need user traffic in order to get revenue; only those articles that were made public and last modified between 1st January 2015 and 31st December 2015 (and still available) were included in the corpus; all articles selected belonged to a specialized section of the newspaper/broadcast websites. As for SciBlogs, only personal blogs that are regularly updated were included in the corpus; all posts that were made public and last modified between 1st January 2015 and 31st December 2015 (and still available) were considered; all posts selected were marked as covering scientific topics through the use of tags. All the texts in SciNews and SciBlogs are automatically tagged for parts-of-speech using the software TagAnt (Anthony 2015). Within the framework of stance evaluation (Hunston 2011; Martin & White 2005) and a rhetorical approach to language (Fahnestock 2011), the analysis can be broken down into four stages: first, identification of trends in the use of evaluative language in news article headlines and blog post titles; second, identification of trends in the use of rhetorical figures in presenting the same scientific event (NASA’s New Horizons mission) over the course of 2015; third, identification of trends in the contextualisation of direct quotes (reporting verbs and accompanying elements) to convey opinion; fourth, identification of linguistic patterns related to selected adverbs used to convey different degrees of confidence in assessing the veracity of a statement.