From postulates formulated by the classics at the turn of the 19th century, to the emergence of sociology of sociology taking place in the early 1970s, sociologists have been seemingly obsessed with finding a general definition of sociology. However, these theories of sociology are often based on lived experience rather than collected data, not seldom focused on drawing boundaries between good and bad sociology, and, unfortunately, sometimes mix up "what sociology is" with "what sociology ought to be". Nevertheless, concepts found within these theories are some of the first disciplinary terms sociology students encounter and continue to form a central part of the jargon used by sociologists to speak about and conceptualize the history of their discipline on an everyday basis. In this presentation, I revisit some of the more famous theories of sociology and pose the question if they can be enhanced by the context provided by empirical studies. The case used for this inquiry is the sociology dissertations defended at the Swedish universities over the last 70 years. The sociology dissertation constitute two corpora – one based on full-texts the other on abstracts – which are scrutinized with qualitative interpretations of models generated by computational text analysis methods. Thus, this is an invitation to consider tools developed during the 21st century within branches of natural language processing and machine learning when undertaking histories of sociology.