Electronic Thesis and Dissertation Repository

Thesis Format

Monograph

Degree

Doctor of Philosophy

Program

Business

Supervisor

Zbaracki, Mark J.

Abstract

This dissertation centers on the discourse concerning the Keystone XL pipeline project. Keystone XL was a proposed mega pipeline that would have transported oil sands from Alberta in Canada to Texas in the United States. Environmental organizations hailed the project’s defeat as a generational climate victory. My thesis offers insights into the compartmentalized discourse over Keystone XL, which pivots on different issues at the local and national levels. At the national level, I observe the now-commonplace polarized dispute over climate action. At the local level, participants sidestepped the topic of climate change, choosing instead to rally around matters of local relevance, which paved the way for the project’s defeat. In my methods section, I investigate the challenges associated with qualitative and topic modeling research when participants refrain from explicit dialogue about a central topic that remains in the background. Furthermore, I expand on the implications of this research to cover any application of Natural Language Processing in mixed-methods research. I emphasize that text data always carry the author’s perspective, and this situatedness necessitates human judgment, even when computer-assisted methods are employed.

Using my mixed-methods approach, I reveal that interactive dynamics contributed to the gap in discourse topics and to the silence on climate change at the local level. Framing is a well-established mechanism; actors engage in anticipatory, defensive framing. That is, they sidestep the topic of climate change to pre-empt pushback. State senators control the discourse over Keystone XL at the local level, and any actor who introduces a non-resonant topic risks swift dismissal. However, the interaction order does not merely permit privileged actors to dismiss others. Through steering, state senators can restrict the topic of conversation and compel other actors to limit their discourse contributions to certain topics. While industry allies made efforts to intervene in favor of Keystone XL, the silence on the topic of climate change and the defeat of Keystone XL based on local concerns were ultimately driven by these interactions at the local level.

Summary for Lay Audience

We often hear about climate change in the media and professional networks like LinkedIn. However, what we hear less about is the gap between these discussions and our lack of progress on the issue. Global emissions continue to increase. In this thesis, I discuss how an alliance of environmental organizations, climate scientists, and local grassroots organizations successfully halted the Keystone XL pipeline project, which is considered one of the most controversial fossil fuel endeavors. Surprisingly, I found that the outcome of the Keystone XL project was not determined by its impact on the climate, but by local issues. In key moments, the actors involved intentionally avoided discussing climate change. To illuminate and utilize this silence as data, I developed a new methodological approach. This approach uses topic modeling, an algorithm that autonomously identifies the topics present in a text. Before applying topic modeling, I independently studied the context, and I used my understanding of the context to deliberately influence the topics generated by the algorithm. By analyzing the data in this sequential manner, I address a significant challenge faced by AI techniques used for textual analysis. These computer-based methods lack the critical judgment of humans and cannot identify important themes unless they are explicit and frequently appear throughout the text data.

Download

Share

COinS