Waning Breakthroughs? Investigating the State of Innovation in the Field of NLP

In this study, we conducted a bibliometric analysis of the Natural Language Processing (NLP) community. Our research hinged on key theoretical frameworks, including Polanyi's republic of science, Kuhn's paradigm shifts, Popper's falsificationism, and Lakatos's scientific research programs. In this study, we hypothesized that authors with higher co-authorship centrality and greater topic diversity could indicate innovative research that contributes to paradigm shifts. Through an evaluation of author productivity, institutional contributions, thematic mapping, topic modeling and co-citation networks, we reveal the vital role of academic institutions, the centrality of specific research themes within the community as well as a shift in research focus from the early pioneering explorations on the foundations of AI to the more recent commercially focused engineering implementations of machine learning techniques. This exploration allowed us to assess the state of the NLP community, and to draw the potential steps necessary to avoid an excessive narrowing of focus in the community's research agendas and to preserve the long-term relevance of NLP-based AI research for the broader scientific community. Our proposed future investigations include expanding the datasets to better identify niche research programs and training deep learning models to analyze semantic content. With this research, we hope to inform researchers, policymakers and business leaders alike on the measures, incentives and research objectives necessary to make the NLP community a true republic of science.