Quest for Sociology : Revisiting Prevailing Understandings of a Discipline with Computational Text Analyses of Dissertations

Abstract: What is sociology? For centuries sociologists have struggled to answer this question and repeatably proclaimed that their discipline is in crisis. The problem has generated a field of its own, the sociology of sociology, where sociologists of knowledge offer concepts for how the paradigmatic status of discipline and its crisis ought to be understood. Yet, the foundation of these understandings has often been limited to conceptual reasonings, historical exposes, and anecdotes from prominent scholars. Following the increasing availability of digitized texts and the development of computational techniques, new venues have been opened for investigating the empirical bearing of what sociology is. This dissertation offers a synthesis of, and a contribution to, this growing literature at the intersection of the sociology of knowledge and computational social science.The starting point is a review of literature in the sociology of sociology that has found that our discipline is believed to exist in a state of fragmentation, lacks a paradigm, and is conditioned by the context of its production. Akin to the supposed crisis, these conceptualizations are often taken for granted rather than being empirically put to test. This is why this dissertation aims to shed new light on the crisis of sociology by empirically scrutinizing prevailing disciplinary understandings with an interpretative and theory-driven methodological approach to computational text analysis (i.e., word correlation networks, topic modeling, stylometry, and shallow neural networks). To account for textual representations of sociological knowledge that are firmly institutionalized and exist across different local contexts, hundreds of dissertations in this discipline published in Sweden between 1980 and 2019 by five main universities have been digitized to form two corpora – 380 full-texts and 850 abstracts. Using these corpora, the conceptualizations are operationalized to be able to scrutinize, and trace, reoccurring instances where dissertations allude to certain images of sociology, which, drawing on the work of Margaret Masterman, can be regarded as crude replicas of paradigms. The study design allows us to problematize prevailing understandings of what sociology is.In contrast to the notion of fragmentation, the corpora are constituted by a core conditioned by local institutions attuned to different paradigmatic images of sociology. A discrepancy is also found between the two corpora where the abstracts appear to follow a divide between qualitative and quantitative research, and the full-texts are characterized by five paradigms with distinct methodological, epistemological, and ontological positions. These results suggest that the coexistence of multiple paradigms has been conflated with fragmentation and that sociologists tend to present their knowledge along the lines of simplified dichotomies. In response to the crisis, a more fruitful approach might be to embrace paradigm pluralism.As a contribution to the sociology of knowledge, this dissertation is an example of how the methodological divide can be overcome by merging insights from the conceptual strand with a hermeneutical take on computational methods to empirically explore taken-for-granted assumptions behind the production of disciplinary knowledge.