An Analysis of Text Functions in Information Visualization

\authororcid Chase Stokes0000-0001-7644-9021, \authororcid Anjana Arunkumar0000-0003-3513-8600, \authororcid Marti A. Hearst0000-0002-4346-1603, and \authororcid Lace Padilla0000-0001-9251-5279 · 2025

Read An Analysis of Text Functions in Information Visualization by \authororcid Chase Stokes0000-0001-7644-9021, \authororcid Anjana Arunkumar0000-0003-3513-8600, \authororcid Marti A. Hearst0000-0002-4346-1603, and \authororcid Lace Padilla0000-0001-9251-5279 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-12334.txt.

Introduction

Chase Stokes and three co-authors build a framework for what text does inside data visualizations. Titles, labels, and notes shape how readers grasp a chart, yet their roles in real designs were poorly mapped. The team coded one hundred twenty real charts and more than eight hundred text elements.

They found ten distinct jobs for text, from naming what maps to what, to adding slanted subtext. Grouping the jobs gave four broad design strategies, such as annotation centered design and narrative framing. Some text even replaces visual elements outright, a role past schemes missed. Titles often do several jobs at once. The framework helps designers use words with intent rather than habit.

Catalog record

CREATOR\authororcid Chase Stokes0000-0001-7644-9021, \authororcid Anjana Arunkumar0000-0003-3513-8600, \authororcid Marti A. Hearst0000-0002-4346-1603, and \authororcid Lace Padilla0000-0001-9251-5279
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE67 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-12334

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Analysis of Text Functions in Information Visualization \authororcid Chase Stokes0000-0001-7644-9021, \authororcid Anjana Arunkumar0000-0003-3513-8600, \authororcid Marti A. Hearst0000-0002-4346-1603, and \authororcid Lace Padilla0000-0001-9251-5279 Abstract Text is an integral but understudied component of visualization design. Although recent studies have examined how text elements (e.g., titles and annotations) influence comprehension, preferences, and predictions, many questions remain about textual design and use in practice. This paper introduces a framework for understanding text functions in information visualizations, building on and filling gaps in prior classifications and taxonomies. Through an analysis of 120 real-world visualizations and 804 text elements, we identified ten distinct text functions, ranging from identifying data mappings to presenting valenced subtext. We further identify patterns in text usage and conduct a factor analysis, revealing four overarching text-informed design strategies: Attribution and Variables, Annotation-Centric Design, Visual Embellishments, and Narrative Framing. In addition to these factors, we explore features of title rhetoric and text multifunctionality, while also uncovering previously unexamined text functions, such as text replacing visual elements. Our findings highlight the flexibility of text, demonstrating how different text elements in a given design can combine to communicate, synthesize, and frame visual information. This framework adds important nuance and detail to existing frameworks that analyze the diverse roles of text in visualization. \onlineid 1611 \vgtccategory Research \authorfooter Chase Stokes and Marti Hearst are with University of California, Berkeley. E-mail: chase_stokes, [email protected] Anjana Arunkumar and Lace Padilla are with Northeastern University. E-mail: a.arunkumar, [email protected] \teaser These designs demonstrate Replace Mappings \xspace : the use of text elements to replace conventional mapping components in a chart, such as Axes \xspace and Legends \xspace . This function showcases the integration of text and visual design choices. Within this paper, we propose and analyze ten distinct text functions, including Replace Mappings \xspace , and show how they can be organized into four factors. i keywords Visualization, text, language, text function, factor analysis, design patterns. 1 Introduction Visualization research has produced important empirical and theoretical work that provides recommendations for using graphical elements (e.g., [ 72 , 15 , 8 , 69 ] ) while historically overlooking the impact of text. Designers use text in almost every visualization, but unlike visual design standards that optimize data encodings for precision and clarity, text-related decisions often rely on designer intuition and conventions. Text plays a critical role in visualizations; titles, captions, and annotations provide context, guide attention and frame understanding. Language is a highly flexible tool, offering countless ways to convey an idea, making it a promising subject for deep analysis. Recent studies have begun to reveal how variations in text placement, function, and style influence reader experience and comprehension (e.g., [ 61 , 60 , 59 , 62 , 75 , 32 , 1 ] ), and prior work has built useful foundations for visualization-based text classifications [ 43 , 52 , 42 , 11 , 3 ] . However, there remain many open questions, including a gap in understanding the specific functions that textual elements perform. Previous efforts often focus on broad categories or isolated uses of text and may not clearly define the specific functions of text elements. The lack of a clear, functional framework creates a barrier to understanding how text works in concert with visual elements to support data communication. We define text function as the specific role a text element plays in supporting the design and data communication of a visualization, focusing on the actions it performs and the target it operates on. The “action” terms borrow verbs directly from prior taxonomies in visualizations [ 52 , 11 ] . The “target” terms were compiled through the exploratory analysis described in Section 3 as well as close readings of taxonomic definition in prior work [ 52 , 43 ] . A function is distinct from the broader notion of the role of text, as studied in prior works [ 59 , 62 , 52 ] , which describes the general purpose or effect of text in guiding reader interpretations. In an effort to further clarify this distinction, we investigated our first research question: RQ1: What are the functions of text in visualization designs? This question seeks to identify and characterize the detailed functions individual text components fulfill. We develop this text function framework via a detailed qualitative analysis of 804 individual text components drawn from 120 visualizations from several domains, such as blogs, government reports, and news sources. In our framework, functions describe how visualizations use language at the level of an individual text component, analyzed independently from but in conjunction with its type (e.g., Titles , Legends , Annotations ). Through the qualitative analysis, we uncover several novel insights into the roles text plays in visualization design. For instance, we find that text usage in visualizations is often multifunctional; Titles can state the chart’s variables, synthesize data, compare values, and hint at emotional valence all at once. Another insight is shown in An Analysis of Text Functions in Information Visualization . Replace Mappings is a previously unexamined function in which a text element replaces a conventional component of a visualization, as when data values on the chart replace the y-axis. The detailed understanding of text function can be used to create stable characterizations of text in visualization design. Based on the foundational understanding of text functions distilled in RQ1 , we explore RQ2: What text design patterns emerge across visualizations, and which functions are engaged in each pattern? We conducted a factor analysis to investigate combinations of text functions within text elements and entire designs, aiming to uncover common design patterns. This analysis revealed four characterizing factors of how text is used across designs, which we named: Attribution and Variables , Annotation-Centric Design , Visual Embellishments , and Narrative Framing . These factors not only highlight trends in text usage but also help clarify how specific text choices can align with broader design goals. Contributions: In this work, we build on existing frameworks for interpreting text in visualizations, focusing on the function of each text component with respect to its content and how it interacts with other elements in the design. We contribute: • A Set of Text Functions : We develop a structured set of text functions, based on prior work and our own detailed analysis. These functions fill existing gaps in text taxonomies. • Application of Text Functions : We apply these function definitions to more than 800 text components using a detailed codebook, achieving high inter-annotator agreement ( Mean=0.97 ). We show how text functions occur across individual text types (e.g., titles), how often a single text element exhibits multiple functions, and other notable uses distinct from overarching design patterns. • Design Patterns Based on Text Function : We examine how text functions are used across text components within a given design by performing a factor analysis on the coded data. We extract four factors that represent the main trends in text design across the corpus. We conduct computational validation of these factors to ensure the robustness and stability of item associations. Clearer guidance on text’s varied roles may help designers use language more effectively. By establishing a structured framework for text functions, we provide a foundation for future improvements to both

Read the complete text