Writing Strategies for Science Communication: Data and Computational Analysis

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 5327–5344, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 5327 Writing Strategies for Science Communication: Data and Computational Analysis Tal August† Lauren Kim† Katharina Reinecke† Noah A. Smith†♦ †Paul G. Allen School of Computer Science & Engineering, University of Washington, Seattle, WA, USA ♦Allen Institute for Artificial Intelligence, Seattle, WA, USA {taugust, lkim143, reinecke, nasmith}@cs.washington.edu Abstract Communicating complex scientific ideas with- out misleading or overwhelming the public is challenging. While science communication guides exist, they rarely offer empirical evi- dence for how their strategies are used in prac- tice. Writing strategies that can be automati- cally recognized could greatly support science communication efforts by enabling tools to de- tect and suggest strategies for writers. We com- pile a set of writing strategies drawn from a wide range of prescriptive sources and develop an annotation scheme allowing humans to rec- ognize them. We collect a corpus of 128K sci- ence writing documents in English and anno- tate a subset of this corpus.1 We use the annota- tions to train transformer-based classifiers and measure the strategies’ use in the larger cor- pus. We find that the use of strategies, such as storytelling and emphasizing the most impor- tant findings, varies significantly across publi- cations with different reader audiences. 1 Introduction Communicating scientific discoveries to a general audience of readers is difficult. A researcher or writer interested in doing so is faced with the chal- lenging task of translating complex scientific ideas in an engaging manner without misleading or over- whelming their audience. There are many guides to science communication (e.g., Blum et al., 2006), but they rarely offer empirical evidence for how their advice is used, or proven effective, in practice. The potential science communicator is then con- fronted with the additional hurdle of understanding how to implement these guidelines in their writing. Effective science communication requires under- standing the unique needs and expectations of dif- ferent audiences and stakeholders in science (Nis- bet and Scheufele, 2009). We envision natural 1Available at https://github.com/talaugust/ scientific-writing-strategies. language processing technologies that help sci- ence writers communicate more effectively. These technologies might automatically classify common strategies in a writer’s own text, support writers to adapt language to specific readers, or guide readers through personalized article recommendations. As a first step, we compile a set of strategies from a wide range of prescriptive science writ- ing sources in English and develop an annotation scheme allowing humans to recognize these strate- gies in texts about science. We introduce a new corpus of 128K university press releases, science blogs, and science magazines and annotate a sub- set of 337 texts. We use the annotations to train transformer-based classifiers to explore the com- municative goals of science writing by analyzing variations in the strategies’ use across several sci- entific communication forums. Our paper is the first computational analysis of writing strategies driven by science communication theory. We find that most strategies are prevalent throughout our corpus and …

Content truncated to 500 words