No abstract

Digital methods are becoming more and more important for text analysis in communications research. However, many computational methods require either relevant technical expertise or multi-disciplinary collaboration, which has impeded their uptake. This article introduces an alternative: computer-assisted linguistic analysis (corpus linguistics), an approach that is increasingly being used outside linguistics and requires less expertise. The article uses a dataset of almost 700 items of health news to demonstrate how such techniques can aid the analysis of (dis)preferred language, sources, stigma and responsibility, framing, and project-specific text analysis. We conclude with an evaluation of the key advantages and limitations of corpus linguistic analysis.

Issue addressed: This study analysed diabetes coverage in 12 Australian metropolitan/national newspapers over a period of 5 years (2013-2017). It aimed to describe quantitative tendencies in diabetes coverage (amount of articles per newspaper and over time) and to identify potential discrepancies between diabetes coverage and societal prevalence of diabetes. The study addressed the following research questions, with a focus on language use: • How frequent are mentions of different types of diabetes? • How are people with diabetes referred to? • How frequent and how distributed are mentions of Aboriginal and Torres Strait Islander people and matters? Methods: Data were collected in electronic format, manually classified and processed using a specialised software program, with a focus on quantitative analysis. Results: 577 articles were classified as news and 117 were classified as "non-news." The Australian Financial Review published the fewest items, followed by the NT News, while the West Australian and the Advertiser published the most. References to "type 2" appear slightly more frequent and more distributed than to "type 1" diabetes. The labelling of people with the noun diabetic/s occurs in about a quarter of the dataset. References to Aboriginal and Torres Strait Islander people or matters appear to be extremely rare in the analysed dataset. Conclusions: Diabetes coverage does not fully align with incidence of diabetes among Australians, and problematic language practices such as the labelling of people as "diabetics" continue to occur. So what? Given the agenda-setting function of the news media, new strategies may be needed to change how Australian metropolitan and national newspapers cover diabetes, especially in relation to incorporating Aboriginal and Torres Strait Islander voices and perspectives.

This study investigates changes in sex education advice from the 1990s to the 2010s. Our research is based on the analysis of an 88,000 word corpus of advice columns from Dolly, a beauty, lifestyle and celebrity magazine aimed at Australian girls. The data are taken from 1994, 1995, 2014 and 2015, with both decades compared against each other to identify any potential shifts in sex education advice. The study combines corpus linguistic techniques with analysis of evaluative language (appraisal). Our analysis reveals a preoccupation with sexual health in the 1990s, shifting to a preoccupation with mental health in the 2010s. We identify a discourse of risk and safety and a discourse of pleasure in the 1990s, and medicalising and normalising discourses of mental health in the 2010s. We also consider interactions between question and answer in the advice pages, to better understand how discourses are introduced and negotiated in such written dialogic texts.

<p>Investigation of local to global scale environmental change is frequently underpinned by data from climate reanalysis products, yet access to these can be challenging for both new and established researchers. The practicalities of working with reanalysis data often includes handling large data files that can place limit users on the scale of analysis they can undertake; and working with specialist data formats (e.g. NetCDF, GRIB) that can pose significant barriers to entry for those who may be unfamiliar with them. Together, these factors are limiting the uptake and application of climate reanalysis data within both research and teaching of environmental science.</p><p>Here we present the Google Earth Engine Climate Tool (GEEClimT), providing an intuitive “point and click” graphical user interface (GUI) for easy extraction of data from 17 climate reanalysis data products relating to atmospheric and oceanic variables (including, but not limited to: ERA5; ERA5-Land; NCEP/NCAR; MERRA; and HYCOM). The GUI is built within the Google Earth Engine geospatial cloud computing platform, meaning users only require an internet connection to rapidly obtain both point data and area averages for user defined regions of interest. To ensure a wide range of usability for researchers, students and instructors, both the GUI and its documentation have been co-created with those who may use reanalysis data for research, teaching, and project purposes. The tool has also been designed with flexibility in mind, allowing it to be easily updated as new datasets become available within the Google Earth Engine data catalogue.</p><p>GEEClimT is shown to allow users with little or no previous experience of working with climate reanalysis data or coding to obtain temporally comprehensive data for their regions and time periods of interest. Case studies demonstrating the application of the tool to different environmental and ecological settings are presented, showcasing its potentially wide applicability to both research and teaching across environmental science.</p>