A suite of qualitative analysis tools
A computational qualitative analysis method for subjectivity-rich interviews.
CISA employs linguistic discourse analysis and narrative-analytic techniques to help the analyst get to the data that matters quickly — and to view that data through a range of analytical lenses. Detection is automated; the analyst can override any classification by hand. Always vanilla. No data leaves your device.
§ 01 · What CISA is
An analytical toolkit that embodies its methodology
In generic qualitative analysis software, the researcher manually applies theoretical categories — coding “tension” or “identity claim” wherever they judge them to occur. The software stores and retrieves what is coded, but imposes no theoretical structure. CISA operates differently: the theoretical categories are constitutive of the instrument itself (Prevett 2026, § 4).
Detection happens automatically, through explicit rule-based linguistic and narrative analysis. The analyst remains in control: every automatic classification can be overridden, reassigned or annotated by hand. CISA reports; the analyst interprets.
Always vanilla
HTML, CSS, JavaScript only. No external libraries, no build step, no dependencies that change beneath you.
Client-side only
All analysis runs inside your browser. Transcripts never travel to a server. Press F12 to verify there are no outbound calls.
No AI at runtime
AI was used as a building partner during development. The tools you use contain none. Rule-based detection only, with full analyst override.
§ 02 · The flagship
CISA Narrative
CISA Narrative reads a transcript and surfaces the moments where the speaker is actively doing identity work — what the tool calls a critical incident segment. It then maps the structural oppositions inside and across those moments, classifies their affective quality, and lays out the narrative arc of the interview.
A short worked example
From the synthetic transcript “Sarah” in the position paper:
I have this terrible fear about money, about not having enough, about spending too much. But actually, the fear makes me more rational. It forces me to plan everything, to use spreadsheets, to track every penny.
The contrastive pivot — but actually — marks a constraint reframed as a resource. Fear is held simultaneously as debilitating and as functional capability. CISA Narrative tags this as a Constraint – Resource opposition with a Managed affective quality, and links it to other moments in the transcript where the same opposition recurs.
Per-transcript outputs
Six of the views CISA Narrative produces from a single transcript.
Interview narrative arc
The narrative structures detected across the interview, in transcript order. The upper band reads the account at a coarse grain, the lower band at a finer one. Grey marks talk where no structure was detected.
Domain arc network
The same reading taken domain by domain. Each node is sized by how much of the account sits in that domain and divided by the structures detected within it. Links mark domains that co-occur in the same incidents.
Interview tensions arc
The same sequence read for tension rather than for structure: all segments in transcript order, coloured by affective quality. The bolt mark indicates a structural opposition.
Domain × Affective Quality
How affective quality distributes across life domains. Each cell is clickable in the tool, revealing the underlying interview talk.
Transcript structure
Inside a detected episode: each stage with the incidents placed there, coloured by the register they are carried in. Stages left empty are shown as empty. Click any stage in the tool to read the incidents beneath it.
Structural opposition co-occurrence
Which opposition types travel together across the transcript. Edge thickness is co-occurrence count; segment slices show affective quality mix.
Dataset and group-level outputs
Within a single browser session, CISA Narrative builds a Session Collection — a working set of analysed transcripts that stays entirely on your device. From there the tool produces a parallel family of cross-participant and dataset-level views.
Cross-Participant Report
Compares patterns across multiple analyses in the Session Collection — structural oppositions held in common, register distributions, transcript-structure shapes. It also compares the narrative arcs of the participants, so that accounts moving through the same structures, and accounts that do not, can be read against one another.
Longitudinal
For datasets with multiple waves: tracks shifts in opposition repertoires and affective quality for the same participant across time points.
Dataset Explorer
Browses the full collection with filters and lenses — participants on one axis, oppositions or registers on another, with every cell traceable to its source.
Participant Grouping
Creates and compares subgroups within the dataset — by demographics, by analytic feature, or by hand — and reports differences between groups on the same metrics.
CISA Discursive Tensions Analysis
An aggregated cross-participant view of where and how discursive tension concentrates across the dataset, including co-occurrence patterns and dominant register signatures.
Export & share
Per-analysis results and whole Session Collections can be exported as structured files for archiving, sharing with collaborators, or moving between machines.
Every output traces back to evidence
Click a cell, a node, or a slice and the underlying interview talk opens beside it. If automatic detection has misread an extract, the analyst can override the classification — reject it, reassign it, or annotate it. CISA reports detection; the analyst remains the interpreter.
§ 03 · Reading identity tensions
Two axes, kept independent
CISA Narrative tracks identity tension along two axes that the framework deliberately keeps separate.
The first axis is what the speaker is working with — the structural opposition they are holding (X versus NOT-X). The second is how they are voicing it — the affective quality of the talk in that moment, ranging across harmonious, easy, managed, stressed, and critical registers. The same opposition can be carried harmoniously in one moment and critically in another; these are distinct discursive positions, not the same content in different moods.
For Sarah, fear is held simultaneously as constraint and as resource — not a contradiction the analyst must resolve, but a structural opposition the speaker is actively living with. CISA Narrative surfaces such moments, classifies the register in which they are voiced, and shows how the same opposition recurs across the interview at different registers. The narrative arcs are calculated separately from this, through the detection of five narrative structures across the sequence of incidents — maintenance, unsettled, transformation, change and fragmentation — so that what a speaker is holding, how they are voicing it, and where the account is travelling remain three separate readings of the same talk.
How to read every panel is set out in the CISA Narrative Learning Guide →
§ 04 · The learning guide
CISA Narrative Learning Guide
A self-paced guide to the method and to the tool, written for researchers rather than for developers. It runs in the browser like everything else here, needs no account, and is free to everyone — whether or not you ever buy the tool.
What it covers
What a critical incident segment is and how detection finds one; the two independent axes; structural oppositions and affective quality; the narrative structures behind the arcs; and how to read every panel the tool produces, panel by panel.
How it is written
Worked examples throughout, using real analytic output rather than idealised diagrams — including the awkward cases, where detection gets something wrong and the analyst has to decide what to do about it.
What it is for
Reading the outputs with enough confidence to defend them: knowing what a figure claims, what it does not claim, and what you would have to check before putting it in a write-up.
Built to be interrogated, not merely read
Every figure in CISA Narrative is designed to be opened. A heatmap cell, a matrix intersection, a node in a network, a stretch of the narrative arc — none of them is a terminus. Each is a way in to the passages beneath it. Three mechanisms are what turn a set of outputs into an analytical instrument.
Click the figure, reach the data
A matrix can tell you that two things co-occur. It cannot tell you what that co-occurrence is, and a number on its own can mislead as easily as it can inform. Selecting a cell opens the incidents that produced it — their affective quality, their domains, and the participant’s actual words.
See the extract in its context
A detected segment is a fragment of a longer account, and a fragment can read very differently once you can see what surrounds it. Roughly a hundred words either side are available at a click — shown for your reading only, never analysed and never counted in any figure.
Change what you disagree with
No template classifies text with complete accuracy, and CISA does not ask you to pretend otherwise. Reject an incident, reassign its structural opposition, or reassign its affective quality. Every control sits beneath the incident it governs and opens on keep, so a change is always deliberate.
§ 05 · The suite
Companion tools and learning resources
CISA Narrative does not stand alone. A short preparation tool gets a transcript into the format the analyser expects, and a learning guide walks newcomers through what the outputs mean.
TranscriptReady
A formatting helper that cleans and standardises interview transcripts before analysis — speaker turns, timestamps, anonymisation placeholders, and the export format CISA Narrative reads.
CISA Narrative Learning Guide
A self-paced, browser-based guide for new users. It walks through detection, the two-axis framework, and how to read every panel the tool produces. Open the guide →
Template Enhancer
An optional tool to calibrate and make an enhanced bespoke template to run in CISA Narrative.
Other qualitative analysis tools
Open Coding and Explore Tool
An open-coding environment with exploratory views for interview data.
Open Coder and Explorer Focus Group Tool
The open coding and exploration approach adapted for multi-speaker focus group data.
TranscriptReady
The same preparation tool serves the open coders: one export format is read by CISA Narrative, the Template Enhancer and the Open Coder alike.
Open Coder Tools Guide
A self-paced guide to open coding with the Open Coder and Explorer — building a code list, coding a transcript, and using the Boolean explorer.
§ 06 · Learning
Interactive Research Methods Learning WebApps
Alongside the CISA tools, a set of browser-based learning WebApps covers postgraduate research methods — research literacy and design, qualitative analysis, and quantitative analysis for small studies. Each one teaches through doing rather than reading, for anyone from a complete beginner to a doctoral student. Like the tools, they are vanilla and client-side, so nothing leaves your device.
§ 07 · The position paper
How CISA was built
The development approach — using AI in the building of the tool but never in its operation — is set out in full in an open-access paper in Forum Qualitative Sozialforschung.
Theory-Operationalising Tool Co-Creation
Prevett, P. (2026). Theory-Operationalising Tool Co-Creation: Building Methodology-Specific Analysis Tools Through Human-AI Collaboration. A Position Statement for Qualitative Research Methodology. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 27(2), Art. 15.
§ 08 · Walkthrough
Short video demo
A guided run-through of CISA Narrative on a short transcript, showing detection, the main panels, and overriding a classification by hand.
Coming soon. A short recorded walkthrough is in preparation.
§ 09 · About and contact
About the developer
Pauline Prevett is a Reader in Education at the University of Manchester, where she leads PhD research training in the School of Environment, Education and Development. She teaches research methods on several postgraduate programmes. Her research focuses on students' financial identity and financial self-efficacy, and on qualitative and mixed methods research methodology.
Get in touch
For licensing enquiries, collaboration, or to learn more about CISA:
pauline.prevett@manchester.ac.uk
§ 10 · QualDeep Insights
QualDeep Insights Ltd
CISA Narrative is a QualDeep Insights Ltd product. Purchase includes the CISA Narrative tool bundled with TranscriptReady, the Template Enhancer, and the CISA Narrative Learning Guide.
The CISA Narrative Learning Guide is also complimentary for everyone.
QualDeep Insights is also open to bespoke tool development — details on request via pauline.prevett@manchester.ac.uk.