Improving Climate Communication through Comprehensive Linguistic Analyses Using Computational Tools

Abstract:

An important lesson on climate communication research is that there is no single way to reach out and inform the public. Different groups conceptualize climate issues in different ways and different groups have different values and assumptions. This variability makes it extremely difficult to effectively and objectively communicate climate information. One of the main challenges is the following: How do we acquire a better understanding of how values and assumptions vary across groups, including political groups? A necessary starting point is to pay close attention to the linguistic content of messages used across current popular media sources. Careful analyses of that information—including how it is realized in language for conservative and progressive media—may ultimately help climate scientists, government agency officials, journalists and others develop more effective messages. Past research has looked at partisan media coverage of climate change, but little attention has been given to the fine-grained linguistic content of such media. And when researchers have done detailed linguistic analyses, they have relied primarily on hand-coding, an approach that is costly, labor intensive, and time-consuming. Our project, building on recent work on partisan news media (Gann & Matlock, 2014; under review) uses high dimensional semantic analyses and other methods of automated classification techniques from the field of natural language processing to quantify how climate issues are characterized in media sources that differ according to political orientation. In addition to discussing varied linguistic patterns, we share new methods for improving climate communication for varied stakeholders, and for developing better assessments of their effectiveness.