I’ve been thinking a lot about how stories can be translated into data visualizations, and what we gain or lose in that process.
As designers, we often listen to stories and consider how to communicate them in ways that help people understand their content, context, and narrative. In a page or poster layout, we establish visual hierarchy, emphasize certain details, and organize information to guide attention, flow, and sequence. In data visualization, we look for patterns and relationships that offer another way to understand the material.
These decisions shape what people see. When we select, sort, and categorize parts of a story, we make some information more visible while allowing other details to recede or disappear. Organizing a story can make it more understandable, but it also changes how that story is experienced and interpreted.
This tension is common in visual design, storytelling, and sensemaking. It raises critical questions about what becomes visible when a narrative is reorganized as data, what gets left out, and how a designer’s choices shape the resulting interpretation.
This tension sparked my curiosity about what happens when a story is translated into data and led me to explore these questions through a personal experiment.
An introduction to my experiment
This year, a redesign of my website prompted me to reassess my research goals and how I describe the connections across my work. The project offered an opportunity to explore my interest in data visualization and the tensions involved in translating a linear narrative into a visual form.
For this experiment, I revised my research statement, translated it into a dataset, and used that dataset to create an interactive visualization. Working with my own narrative allowed me to examine both my research and the decisions involved in representing a story as data.
From narrative to a visual network
I began with my research statement, a linear narrative that spans my homepage introduction, research overview, and three detailed sections of Community-based Practices, Interactive Experiences, and Experimental Making.
I retained the full narrative in the workbook so I could return to its context throughout the process. The frequency analysis counted 1,071 words, including section headings but excluding navigation labels, and identified 385 distinct terms. These counts describe the text used for analysis.
Identifying word frequency
I ran the 1,071-word narrative through a frequency checker, following a process similar to the one Camila Afanador-Llach and I used in Mapping Communication Design Research Trends Through Data Visualization.
Original narrative: the research statement as a reference for the analysis.
Identifying distinct phrases
After reviewing individual words, my next step was to identify words that were meant to be combined as phrases, but were separated as part of the frequency analysis process.
I reviewed the text for words that belonged together as meaningful units. Phrases such as creative technologies, critical reflection, data visualizations, and digital interfaces described more complete concepts than their component words could convey separately. I selected phrases because they named an idea together, rather than simply because they appeared next to one another.
Combining 69 selected phrases produced 408 distinct words and phrases. The distinct count increased from 385 because phrases became new terms while some component words also remained as standalone entries.
Identifying omissions
I then reviewed which terms would not contribute to a concept vocabulary and therefore could be omitted. Grammatical words, broad descriptors, and other terms without a sufficiently specific conceptual role were marked for omission.
I also set aside verbs such as function, gather, and identify. They did not name concepts on their own, but could help describe relationships later.
Suggested omissions: reviewing grammatical and broad terms that do not contribute to the concept vocabulary.
These choices show how a designer’s decisions influence which meanings remain visible as a narrative changes form.
For example, I retained listening as a concept while treating people as too broad on its own. I omitted standalone design while retaining more specific phrases such as interactive design.
This review left 219 distinct words and phrases.
Building a framework: from words to concept groups
The 219 retained words and phrases were still too many to use individually in a network diagram. I grouped related terms to create a smaller list of concepts.
I added a concept-group column beside the original terms and counts so each grouping could be traced back to its source language. I often chose an existing term as the group name, selecting the one that most clearly communicated the shared concept.
For example, communities, community, community engagement, community recruitment, community-based practices, community-centered interventions, community-led initiatives, and public-facing were all grouped under Community Engagement. Eight distinct words and phrases, appearing fourteen times in the narrative, contributed to one concept group. This was an interpretive decision about how the language connected across the narrative, with the original terms and counts preserved in the spreadsheet.
I refined these assignments through several rounds of review. For example, I kept Play as a distinct concept after considering whether to group it with Experimentation. Either choice was possible, but keeping it separate gave it greater visibility. By contrast, I grouped power and decision-making with Shared Authorship, treating them as dimensions of that concept. These decisions show how designers shape what remains visible, what is combined, and what is left out. My choices reflected how I understood the terms within my practice and narrative, including whether each needed its own label or could be meaningfully represented within a broader concept.
Before finalizing this phase of work, I also strategically chose not to use Community-based Practices, Interactive Experiences, and Experimental Making as concept groups. These headings already structured my linear research statement, and I wanted a emphasize different concept groups that could help me see connections across those areas in different ways.
Ultimately, I assigned 174 of the 219 retained terms to 40 concept groups. The remaining 45 terms did not fit within those groups and were left out, but they did help inform the categories that would follow next.
Organizing concept groups into four categories
Finally, I organized the 40 concept groups into four categories. Some terms that I had left outside the concept groups, including research, approaches, processes, forms, and outcomes, helped me identify broader ways to organize the concepts and develop this framework.
- Research Materials: What the research engages, examines, or draws upon.
- Approaches/Process: How the research and design work is conducted.
- Forms: What is designed, produced, or experienced.
- Outcomes: What the work supports or makes possible.
Concept groupings fit within the categorical framework: Research, Approaches, Forms, Outcomes
Creating connections: developing relationships
I now had a manageable vocabulary organized within a framework that I could start to visualize. The concepts would become nodes in the network diagram. The next step was to define the relationships represented by the links between them.
I began by considering connections across categories. For example, Oral Histories, a research material, contribute to Storytelling, a form. Systems Thinking, an approach, supports Sensemaking, an outcome. I drew on the verbs set aside during the omission review to articulate these connections.
Developing relationships: verbs connect pairs of concepts in readable statements.
At this stage, I did not return to the original narrative to extract relationships. I authored connections informed by my research practice, using the concept vocabulary as a starting point. Each connection required an interpretive rationale. This marked a shift from analyzing the language of the statement to proposing relationships among its concepts.
The working relationship list included every concept in at least one connection. I limited this experiment to a selected set of relationships rather than attempting to map every possible connection.
Developing unique pathways
I also experimented with developing conceptual pathways. I thought about specific projects that I have done or how I ground my research to create these pathways. This was a way to show more complicated relationships between concept nodes.
I developed a small set of pathways for this experiment. They offered a way to explore connections among several concepts at once, with room to develop further pathways later.
Coding the interactive network
I used the concept, relationship, and pathway data to develop an interactive network for my website. I built the visualization with HTML, CSS, and JavaScript, working with AI to help write and revise the code.
Concepts became nodes, and the relationship statements provided the basis for the links between them. The four categories organized the concepts, while selected pathways offered ways to explore connections across those categories.
Moving into an interactive format changes the way readers navigate terms and connections within a text, which influences how the research statement is understood. Readers can begin with a concept that interests them and follow its connections, working with a text in a non-linear format, rather than reading something in sequence. The highlighted pathways provide selected routes through the material, while leaving room to explore beyond them.
My Learnings: What changed in the translation?
Each stage changed what the narrative made visible. The frequency list brought recurring language into view, but separated words from the sentences that explained their meaning. Phrase pairing preserved some of that specificity by treating expressions such as oral histories and shared authorship as complete units.
Grouping terms into concept groups introduced further compression. Community recruitment and community-led initiatives, for example, contributed to Community Engagement, even though they describe different activities and different degrees of community agency. The broader label made a shared concern easier to recognize, while making those distinctions less visible in the network.
Omission also changed the emphasis. Removing standalone design while retaining interactive design and other specific phrases prioritized particular forms of practice over a broad professional term. Separating relationship verbs from concept terms and reassigning them gave the concept nodes new meaning and definitions. From a personal point of view, this analytical process helped me understand more deeply what these concept terms meant and how I was using them in my practice. This exercise gave me a much stronger understanding of what my writing was about.
The four categories also offered another helpful way to organize the statement. Research materials, approaches and processes, forms, and outcomes allowed concepts to be considered by their role in the work, rather than only within the three areas that structured my original narrative. Writing relationships and selecting pathways required me to make those connections explicit.
The experiment helped me describe my research through the relationships among its materials, methods, forms, and outcomes. It also made the decisions involved in that description more visible to me. The network offers one interpretation of my practice, while the narrative and spreadsheet preserve context that the visualization cannot fully convey.
I invite you to explore the network above and perhaps try this experiment with your own research statement or another piece of writing you’d like to examine from a different perspective. Let me know what insights emerge and whether the process helps you see new connections in your work.