Identifying social science engagement within agroecology: Classifying transdisciplinary literature with a semi-automated textual classification method

Abstract

Interdisciplinary and transdisciplinary fields of inquiry and action have been important academic frontiers in recent years. The field of agroecology is a prime example of transdisciplinarity. With roots in the biophysical sciences, social sciences, and peasant movements, publications in agroecology have been growing rapidly in recent decades. Here we explain a method-the script-expert adaptive classification (SEAC) method-that allows us to examine the engagements between agroecology and the social sciences by identifying publications within the agroecological literature that engage with social science at various levels. Using the term "agroecology" and its iterations, we gathered a corpus of agroecology literature up to and including 2019 with 12,398 unique publications from five publication databases-Scopus, Web of Science, Agricola, CAB Direct, and EconLit. Using the SEAC method we then classified each publication as engaged, partially engaged, and not engaged with social sciences and separated this Agroecology Corpus 2019 into three corpora: agroecology engaged with social sciences (with 3,125 publications), agroecology not engaged with social sciences (with 7,039 publications), and agroecology with uncertain engagement with social science (with 2,234 publications) or unclassifiable. This article explains the SEAC method in detail so other transdisciplinary scholars can replicate and/or adapt it for similar purposes. We also assess the SEAC method's value in identifying social science publications relative to the classification systems of the major multidisciplinary bibliographic databases, Scopus, and Web of Science. We conclude by discussing the strengths and weaknesses of the SEAC method and by pointing to further questions about agroecology and the social sciences to be asked of the corpora.