Sentiment analysis as a measure of conservation culture in scientific literature

Abstract

Culturomics is emerging as an important field within science, as a way to measure attitudes and beliefs and their dynamics across time and space via quantitative analysis of digitized data from literature, news, film, social media, and more. Sentiment analysis is a culturomics tool that, within the last decade, has provided a means to quantify the polarity of attitudes expressed within various media. Conservation science is a crisis discipline; therefore, accurate and effective communication are paramount. We investigated how conservation scientists communicate their findings through scientific journal articles. We analyzed 15,001 abstracts from articles published from 1998 to 2017 in 6 conservation‐focused journals selected based on indexing in scientific databases. Articles were categorized by year, focal taxa, and the conservation status of the focal species. We calculated mean sentiment score for each abstract (mean adjusted z score) based on 4 lexicons (Jockers‐Rinker, National Research Council, Bing, and AFINN). We found a significant positive annual trend in the sentiment scores of articles. We also observed a significant trend toward increasing negativity along the spectrum of conservation status categories (i.e., from least concern to extinct). There were some clear differences in the sentiments with which research on different taxa was reported, however. For example, abstracts mentioning lobe finned fishes tended to have high sentiment scores, which could be related to the rediscovery of the coelacanth driving a positive narrative. Contrastingly, abstracts mentioning elasmobranchs had low scores, possibly reflecting the negative sentiment score associated with the word shark. Sentiment analysis has applications in science, especially as it pertains to conservation psychology, and we suggest a new science‐based lexicon be developed specifically for the field of conservation.

Digital data sources and methods for conservation culturomics

“…These include recognizing conservationoriented constituencies and promoting public understanding of conservation issues (Ladle et al 2016). For example, exploring and shaping the evolution of conservation culture (Lennox et al 2020), particularly outside of the academic environment, may be a subject worthy of further exploration with culturomics approaches. In-depth exploration of these topics in conservation would greatly benefit from expertise in areas such as cultural evolution, digital humanities, media studies, social marketing, linguistics, and psychology.…”

Invasion Culturomics and iEcology

“…Researchers use R-based user-developed packages to conduct sentiment analyses (Lennox et al 2020;Correia et al 2021). These analyses quantify polarity of attitudes expressed in texts by assigning sentiment values to text strings based on algorithms and established lexicons (Lennox et al 2020). Sentiment analysis provides an inexpensive metric to examine public perceptions or attitudes, for example, in support or opposition to management measures (Fig.…”

classecol: Classifiers to understand public opinions of nature

“…Furthermore, in some lexicons, species names can have negative scores (e.g. 'shark') which can bias results if we are interested in human-nature relationships (Lennox, Veríssimo, Twardek, Davis, & Jarić, 2020).…”