Critical Incident Semantic Analysis (CISA): Advancing Narrative and Identity Research in Qualitative Studies with Digital Tools

Abstract

This panel discussion introduces Critical Incident Semantic Analysis (CISA), an interpretivist computational narrative methodology designed to analyse subjectivity and identity work through transcripts of interview-generated discourse. CISA represents an effort to bridge qualitative interpretive traditions and the emerging landscape of digital analytical tools, by building digital supports that operationalise theoretical sophistication without reducing human interpretation to algorithmic outputs. It aims to foster discussion on developing bespoke, transparent, and ethically responsible digital analysis tools that enhance researchers’ interpretive capacity while honouring the epistemological commitments of qualitative inquiry. 

CISA was developed to address three core challenges faced by qualitative researchers: first, the time-intensive nature of traditional manual coding when handling large, loosely structured datasets;  second, the challenge of systematic cross-case comparison with loosely structured interview transcripts, and third, the lack of theoretically informed computational methods capable of capturing the subtleties of narrative meaning and identity construction. In response, CISA operationalises how individuals construct meaning about their identities by detecting critical moments that reveal key identity tensions and shifts. These "critical incidents" are moments where language, emotion, or perspective signal a reconfiguration of self-understanding, providing deep insight into identity processes. CISA’s analytical engine uses semantic recognition and structural patterning to highlight such incidents.

CISA’s theoretical foundation integrates Bruner’s (1990) narrative construction of self, Bakhtin’s (1981) concept of multi-voicedness, and Harré’s (1983) positioning theory, all of which conceive identity as a dialogic, contextually embedded phenomenon. Building on King’s (2012) template analysis, CISA advances this lineage by embedding theoretical templates directly into the tool’s analytical logic. The methodology applies both a situated belief template—representing the individual’s local meaning system—and a narrative structure template that traces transformations over time. Together, these templates allow the analysis to move beyond surface-level coding into interpreting contradictions, transitions, and points of narrative tension that drive identity movement. In doing so, CISA produces a structured representation of participants’ identity journeys, linking micro-level discourse to broader social positioning.

CISA’s digital tools are entirely client-side, developed in vanilla JavaScript and embedded within HTML. This ensures that all data processing occurs locally in the user’s browser, protecting data confidentiality and meeting strict university ethics board requirements. Unlike AI-assisted analysis tools, CISA deliberately excludes server-side, AI machine learning or cloud processing, providing a model for ethical, privacy-respecting digital research infrastructure that comply with potential data security concerns raided by university ethics committees. This approach aligns with the growing need for transparent, auditable computational tools in qualitative research, particularly as debates about AI’s ethical implications intensify. CISA tools demonstrate its computational functions are designed to assist rather than replace researchers’ interpretive role. Automated pattern detection identifies potential areas of significance, but ultimate analytic judgment remains with the researcher, who validates or rejects suggestions based on theoretical reasoning. This design explicitly positions automation as a collaborator within an interpretivist workflow as a “second very systematic reader” rather than an authority.

The CISA tool suite currently includes three main digital tool applications: DETECT, MAPLED, and NARRATIVE, each serving complementary analytical purposes. This modular design reflects CISA’s principle of analytical complementarity, each tool can function independently, yet together they provide a holistic framework for studying identity as dynamic, situated, and discursively constituted.

CISA’s detection system builds upon a double binary vector detection model, which allows simultaneous analysis of textual and semantic dimensions, so that positioning around situated constructs or beliefs can be captured and traced through the transcript text. This dual-layer design improves accuracy in identifying critical incidents without necessarily relying on opaque black-box algorithms. Such methodological transparency is central to CISA’s ethos, as it ensures that the theoretical rationale behind analytic operations is visible, reviewable, and open to scholarly debate.

From a technical perspective, CISA demonstrates how narrative constructions can be encoded into digital logic. For example, rather than quantifying word frequencies or clustering by mathematical proximity alone, CISA’s semantic detection algorithms are informed by narrative principles, such as shifts in tense, modality, and affect, that signal identity tension. This embedding of theory into computation challenges the false dichotomy between qualitative richness and computational structure, showing that interpretive reasoning can coexist with systematic digital processing.

The development process itself followed a reflexive, iterative cycle, combining methodological design with user testing and pedagogical integration. Each iteration refined not only the tool’s interface and functionality but also its epistemological coherence, ensuring that digital affordances served rather than distorted interpretive practice. The resulting suite promotes both analytical efficiency and methodological rigour, allowing researchers to manage large qualitative datasets without sacrificing narrative integrity or theoretical sensitivity.

Ethical and pedagogical considerations have been central throughout CISA’s development. Given the increasing demand for digital literacy among qualitative researchers, CISA includes an independent learning resource and demonstration guide to accompany the tool suite. These resources introduce users to both the theoretical underpinnings and the interpretive logic of the software, helping ensure that automation enhances rather than overshadows human judgment.

Practically, CISA has already demonstrated significant potential in educational and identity research contexts, where participants’ narratives often express complex negotiations of self, belonging, and transformation. For instance, studies of academic identity formation, professional transition, or intercultural adaptation can benefit from CISA’s capacity to identify critical junctures in narrative meaning-making. The tool’s flexibility also allows adaptation to other domains—such as health, organisational, or community research—where identity and voice are central analytical foci. By offering researchers a systematic yet interpretive means of identifying and visualising narrative change, CISA bridges a long-standing gap between human interpretation and computational analytic support.

In the broader context of qualitative digital innovation, CISA contributes to an emerging movement of researcher-built, theory-driven digital tools that differ fundamentally from commercial “AI assistants” or generic NLP platforms. These bespoke tools foreground interpretive responsibility, methodological transparency, and researcher reflexivity, principles that are often sidelined in automated analytics. CISA’s development exemplifies how qualitative researchers can engage proactively with technology, shaping it to serve disciplinary values rather than being shaped by technological determinism.

During this panel, attendees will engage both theoretically and practically with the CISA methodology. Presentations will alternate between conceptual exposition and live demonstration, followed by a participatory discussion exploring how such tools might influence the future of qualitative research. The presenters will provide a guided demonstration using synthetic data, receive a QR code to access the demo guide, and be invited to critically reflect on their own methodological needs and the ethical parameters of digital tool design.

Ultimately, this panel aims to initiate a community conversation around how interpretive computation can expand the possibilities of qualitative inquiry while preserving its human-centred foundations. The panel will encourage dialogue about how to make tools that embody, not dilute, the principles of qualitative analysis. In doing so, it suggests CISA not merely as a methodology or software suite, but as a conceptual and ethical framework for future digital qualitative innovation.