Beyond Methodological Boundaries: Histories and Tools for Text-Based Social Research

Abstract

Methodological practices in the social sciences are shaped not only by substantive research questions but also by inherited research traditions that organize training, standards of evaluation, and the justification of method choice. In text-based research, these traditions have often developed in parallel despite their shared object of inquiry, contributing to recurring tensions around interpretation, inference, and scale. This dissertation examines how such methodological boundaries are historically produced and rhetorically sustained, and explores alternative designs for text-based social research that begin from epistemic aims rather than inherited methodological allegiances. The first chapter offers a genealogy of philosophical framing in education methodological debates, focusing on the emergence and persistence of the “paradigmatic view,” according to which qualitative and quantitative methods rest on incommensurable philosophical foundations. I argue that this view emerged partly in response to the marginalization of qualitative research, but came to function rhetorically as a strategy for constructing methodological enclaves, insulating positions from criticism, and fragmenting the field. The chapter also identifies mutually reinforcing mechanisms that continue to reproduce and stabilize this view over time. The second and third chapters develop constructive alternatives for text analysis, each introducing a new framework alongside open-source tools for applying it in practice. Chapter 2 introduces the Latent Interview Protocol Engineer (LIPE), a framework for analyzing interview corpora by recovering the latent interview protocol from interviewer questions as enacted in practice. Designed to support qualitative inquiry at scale, LIPE uses computational techniques to assist with organization and retrieval while preserving interpretive judgment and contextual sensitivity. Chapter 3 introduces Hierarchical Propositions (HiPr), a framework for representing texts as hierarchical propositions. By treating propositions as flexible bundles of semantic-role features and organizing them through entailment relations, HiPr enables researchers to recover semantically richer and more interpretable features than word-based models while avoiding the opacity of embedding-based representations in exploratory research. These representations support downstream statistical analysis, hierarchical exploration of large corpora, and semantically richer retrieval. Taken together, these chapters argue that methodological commitments are not merely technical choices but constitutive features of research practice. By interrogating inherited methodological dispositions and demonstrating concrete alternatives, the dissertation advances a more reflexive and integrative approach to text-based social research—one that combines interpretive, inferential, and computational resources without reducing one to another.