Content, Contribution, and Knowledge Consumption: Uncovering Hidden Topic Structure and Rhetorical Signals in Scientific Texts
Antons, David
Content, Contribution, and Knowledge Consumption: Uncovering Hidden Topic Structure and Rhetorical Signals in Scientific Texts - 3035-3076 p.
Knowledge production and scientific discourse are observable in published scholarly texts. Citations capture knowledge consumption and impact. Drawing from the sociology of science, our theoretical framework posits scientific communities as thought collectives with distinctive thought styles that embed a hidden topic structure and rhetorical signals into a journal’s published articles. We hypothesize and uncover how an article’s topic attributes (structure, focus, and newness) and rhetorical attributes (inclusiveness, exclusiveness, tentativeness, and certainty) are related to future knowledge consumption. We empirically test our ideas by applying text mining algorithms to model topics and extract rhetorical signals from 1,646 strategy articles composed of nearly 18 million words generating 172,237 citations over 35 years. We find that strategy articles’ hidden topic structure explains 14% of variance in scientific impact. We also show that topic focus and topic newness each independently, directly, and significantly increase impact. As for newness, the first two articles published on a new topic each generate a citation premium >100%, which is higher within the focal thought collective than outside. Importantly, we find that the citation premium of newness increases with greater topic focus (which attracts attention) and greater inflow of prior intracollective knowledge (which enhances absorption). Impact also increases when authors present new topics using a rhetorical style that is more tentative than certain. Overall, our findings demonstrate that topic and rhetorical attributes, as constitutive elements of scientific content, are independently and interdependently related to the consumption of strategy articles across thought collectives in management research.
scientific impact management research citation analysis text mining latent Dirichlet allocation
Content, Contribution, and Knowledge Consumption: Uncovering Hidden Topic Structure and Rhetorical Signals in Scientific Texts - 3035-3076 p.
Knowledge production and scientific discourse are observable in published scholarly texts. Citations capture knowledge consumption and impact. Drawing from the sociology of science, our theoretical framework posits scientific communities as thought collectives with distinctive thought styles that embed a hidden topic structure and rhetorical signals into a journal’s published articles. We hypothesize and uncover how an article’s topic attributes (structure, focus, and newness) and rhetorical attributes (inclusiveness, exclusiveness, tentativeness, and certainty) are related to future knowledge consumption. We empirically test our ideas by applying text mining algorithms to model topics and extract rhetorical signals from 1,646 strategy articles composed of nearly 18 million words generating 172,237 citations over 35 years. We find that strategy articles’ hidden topic structure explains 14% of variance in scientific impact. We also show that topic focus and topic newness each independently, directly, and significantly increase impact. As for newness, the first two articles published on a new topic each generate a citation premium >100%, which is higher within the focal thought collective than outside. Importantly, we find that the citation premium of newness increases with greater topic focus (which attracts attention) and greater inflow of prior intracollective knowledge (which enhances absorption). Impact also increases when authors present new topics using a rhetorical style that is more tentative than certain. Overall, our findings demonstrate that topic and rhetorical attributes, as constitutive elements of scientific content, are independently and interdependently related to the consumption of strategy articles across thought collectives in management research.
scientific impact management research citation analysis text mining latent Dirichlet allocation