What ‘Work’ Can Dataviz Do in Popular Science Communication?

Martin Engebretsen

Department of Nordic and Media Studies, University of Agder, Universitetsveien 25, Kristiansand, Norway

Keywords: Data Visualization, Popular Science Communication, Social Semiotics, Discourse Analysis, Design Practices

Abstract: Data visualizations have proliferated on public arenas for information and communication – in journalism,

PR and governmental information as well as in popular science communication (PSC). In the existing

literature on PSC, simplicity, relevance and trust are identified as critical factors for the communication to

succeed. This position paper argue that data visualizations represent a semiotic resource with unique potentials

regarding all of these criteria. The paper aims at presenting a theoretical and methodological framework for

studying data visualizations applied in popular science discourses. The main goals are a) to introduce social

semiotics as an advanced analytical tool for the scrutiny of data visualizations, b) to introduce PUS (Public

understanding of science) as a field relevant for empirical studies of data visualization, and c) to present a

method of analysis combining a small scale corpus analysis with multimodal close reading of selected

visualizations.

1 INTRODUCTION

Data visualization (DV) represents a form of visual

communication well integrated in scientific

discourses. In disciplines where quantitative data play

a central role – particularly in the natural and social

sciences – graphs, charts and maps represent an

irreplaceable semiotic resource for revealing patterns

and relations in large amount of data (Tufte, 2001;

Few, 2012). In the professional scientific discourses,

the codes and conventions related to the production

and interpretation of DVs are strong, and normally

shared by the producer and the reader. In public

dissemination of science, on the other hand, both

verbal and visual forms of communication need to be

adapted to the needs of non-scientists, who share

neither verbal terminologies with the scientists, nor

their forms of visual-numeric literacy. In this paper

we ask: What role do DVs play in popular science

communication (PSC)? What are their semiotic and

social functions in the discourse? And how do they

apply to the norms and expectations of the genre in

which they are embedded?

https://orcid.org/0000-0001-8002-1423

See: https://www.malofiejgraphics.com/;

https://datajournalismawards.org/ and

https://www.informationisbeautifulawards.com/

During the latest decades, data visualizations

have proliferated on public arenas for information and

communication – in journalism, PR, governmental

information etc. (cf. Cairo, 2013; Kennedy et al.,

2016a and b; Engebretsen, 2017). This development

is driven by, among other factors, a growing access to

data from a variety of sources and a rapid

development of cheap and easy-to-use visualization

tools (Engebretsen et al., 2018). However, on these

arenas, DVs meet a heterogenous group of readers,

representing big variations concerning visual and

numeric literacy (D’Ignazio & Klein, 2016; Pinney,

Forthcoming). Advanced DVs, with complex visual

codes and high density of variables and values, face a

high risk of being misunderstood or neglected by

members of the general public. On the other hand, a

creative, surprising and visually attractive DV may

receive a lot of attention and stimulate the motivation

for further reading (cf. Cairo, 2016; Allen, 2018).

What is regarded as best practices of data

visualization in the public, is celebrated in awards like

the Malofiej Award, the Data Journalism Award, and

the Kantar Information is Beautiful Award

260

Engebretsen, M.

What ‘Work’ Can Dataviz Do in Popular Science Communication?.

DOI: 10.5220/0009094002600264

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 3: IVAPP, pages

260-264

ISBN: 978-989-758-402-2; ISSN: 2184-4321

In the existing literature on public understanding

of science (PUS), certain critical factors are identified

regarding popular science communication. These

values relate to simplicity, relevance and trust (cf.

Bauer, 2009). Complex scientific problems and

findings need to be simplified, in order to be

understood by non-scientists. Further, these findings

need to be contextualized and related to issues

important to the general reader, in order to gain

attention and emotional engagement. And, finally, the

chosen forms of communication need to evoke trust,

both in the communicator – be it a news organization

or a professional mediator – and in the scientific

source behind the popularized presentation. The

argument underlying this paper, is that data

visualizations represent a text type with particular

affordances regarding both simplicity, relevance and

trust. Simplification of complexity follows with the

reductional nature of DVs, being the result of a

number of choices made by the producer through the

pipeline of production. Relevance and emotional

engagement can be evoked by the message carried by

the data itself, for instance when a chart shows the

development of criminal incidents in a specific urban

area. It can also be evoked by the contextualization

provided by the verbal co-text, or by the interactive

options offered in certain digital contexts. Finally;

trust can be built through transparency regarding the

production process; opening the “black box”. Such

transparency can be achieved by informing broadly

about the scientific sources and the data sources

behind the DV, the methods used in the production of

it, as well as the institution and the persons behind the

popular version (Engebretsen, 2017).

This position paper aims at presenting a

theoretical and methodological framework for

studying DVs in popular science discourses. The

main goals are a) to introduce social semiotics as an

advanced analytical tool for the scrutiny of data

visualization as a multimodal text type situated in

particular social practices, b) to introduce PUS

(public understanding of science) as a field relevant

for empirical, discourse oriented studies of data

visualization, and c) to present a method of analysis

combining a small scale corpus analysis with

multimodal close reading of selected DVs. The

results of the study will be particularly relevant for

science journalists and others working with science

communication, as well as students and researchers in

the field of information visualization and the field of

Science and Technology in Society (STS).

2 THEORETICAL FRAMEWORK

The main framework for the planned study is social

semiotic discourse analysis (SSDA). This approach to

texts and other meaningful cultural artefacts can be

traced back to the Australian linguist Michael

Halliday (Halliday, 1978). His theories of meaning-

making through verbal utterances have been adapted

to multimodal and mediated contexts, e.g. by Gunther

Kress and Theo van Leeuwen (Kress and van

Leeuwen, 2006; van Leeuwen, 2005). The study is

also inspired by Norman Fairclough's model of

critical discourse analysis (Fairclough, 2012), aiming

at revealing the social and political impact of specific

textual choices. This theoretical framework invites an

analytical approach first focusing on semiotic,

meaning-making structures in the visualization itself,

and then on the role that the DV has in the textual

whole in which the DV is embedded. Finally, the

findings on these semiotic-technological levels are

discussed with regards to the norms and conventions

of the specific genre and to the socio-cultural practice

in which the DV is an integrated element (see more

about methodology below).

A core concept in social semiotic theory, is that of

the three metafunctions. The metafunctions refer to

the ways in which an expression relates to different

aspects of the context: to the experienced world, to

the participants in the discourse and to the textual and

technological resources applied in the production of

the message. These metafunctions correspond to three

types of meaning. Experiental meaning concerns

what the text can say about (an aspect of) the world.

Inter-personal meaning concerns the construction of

social relations between the participants in the

discourse – e.g. between the producer and the reader

of a DV as well as between the reader and the

persons/groups represented by the DV. Finally,

compositional meaning concerns the ways in which

semiotic and media-technological resources are used

to create wholeness and coherence in the textual

output (Halliday, 1978; Kress & van Leeuwen, 2006).

All of these three dimensions of meaning are realized

as semiotic potentials (although not necessarily

identified by all readers), in any instance of

meaningful expression – be it a 5-word e-message or

a 50-minute documentary film – but they are

obviously realized in very different ways. A relevant

question is thus; how are the three metafunctions

realized in DVs applied in public science

communication, and how do they contribute to

understanding, engagement and trust?

In a similar study conducted in 2017, the

framework of SSDA was applied in an analysis of 17

What ‘Work’ Can Dataviz Do in Popular Science Communication?

261

journalistic data visualizations, collected from four

major Norwegian news sites (Engebretsen, 2017).

The study revealed that many of the DVs were

designed in a way that made complex patterns in the

data material easy to perceive, and thus supported the

processing of experiental meaning. Interpersonal

meaning potentials were, on the other hand, less

focused. Only eight of the 17 DVs provided any

information about the producers behind the

visualizations, and only four offered substantial

information regarding the methods used in the

process of production. 10 of the DVs offered elements

of interactivity, inviting the readers to explore the

underlying data by themselves. This study provides a

relevant model for analysis, although the DVs were

collected from a different domain of public discourse

than the PSC-discourse described in this position

paper.

The other part of the theoretical framework for the

study, is that of public understanding of science

(PUS). PUS is a field of activity as well as an area of

social research, closely related to the wider field of

Science, Technology and Society (STS). PUS-related

studies include the building of theory and models,

qualitative case studies as well as surveys and other

quantitative studies of science communication taking

place in a range of public genres; science blogs,

science journalism, popular science magazines,

museum exhibitions etc. The PUS-discourse has

historically, according to Bauer (2009), focused on

three different problems. In the 60s and 70s, the focus

was on a deficit of knowledge in the general audience.

In the 80s, the dominating concern was a deficit of

attention, interest and support of science in the

general public. Since the 90s, much attention in the

PUS-discourse is given to the lack of trust to science

as well as to the media. The discourse of fake news

(having grown in intensity in the Trump-period), is a

symptom (or a driver) of the deficit of trust to the

media. To illustrate the issue of low trust in science,

one can look to Norway, a nation with a high

educational attainment,

yet, with a very high density

of climate skeptics. In the period 2013-2018, between

22 and 27 per cent of the adult Norwegian population

were skeptical to the idea that climate changes are

related to human activities, in spite of what is

massively communicated by scientists.

Modelling the interaction between the “esoteric”

scientific discourses going on among scientists, and

the “exoteric” discourses of science on the public and

According to National Statistics Institute of Norway, 34

per cent of the Norwegian population have education on

a bachelor’s level or higher. https://www.ssb.

no/en/utdanning/statistikker/utniv

private arenas, Bauer – referring to Flecks core-

periphery-model from 1937 (see Fleck, 1979) – states

that the further away from the esoteric core, the

exoteric discourses are characterized by a growing

gradient of simplification, concreteness and certainty

of judgement. In other words; in a highly popularized

presentation of a scientific result, one must expect to

find a higher degree of simplification, of visual

illustration and of certainty (i.e. a lack of reservations

and modifications) than what is expected in e.g. a

textbook in higher education. Bauer calls for more

discourse-oriented studies – as a complement to the

far more frequent quantitative studies – in future

investigations of the PUS dynamics, where the

relationships between the esoteric and the (different

levels of) exoteric discourses of science ought to be

closely investigated.

Some commentators are less concerned about

public deficits regarding knowledge, interest and trust

in their approach to science communication, and

more concerned about dialogue and active

participation. They call for a less top-down and more

interactive, mutual and dialogical view on the

interaction between scientists, science

communicators and the public audience (Riise, 2008;

Santerre, 2008). Davies & Horst (2016) model

science communication as a non-hierarchic eco-

system, and point to its large complexity of actors,

epistemologies and discursive elements. In a

dialogical, non-hierarchic approach to science

communication, the style of expression and the inter-

personal dimension of meaning making play a

substantial role in the construction of the participants’

identities and their discursive roles and power.

In the intersection between these two

frameworks, where the analytical tools of social

semiotic theory are focused by core issues in the field

of PUS, we can extract a more nuanced set of research

questions in our study of data visualization in science

communication, building on the broad and general

questions formulated in the initial paragraph. We now

ask:

• What characterizes the visual codes applied in

DVs in successful PSC to inform about aspects of

the world?

o What DV types are used? What visual codes,

metaphors, forms and colors are applied?

What is the level of information density?

https://www.bt.no/btmeninger/debatt/i/LALy5V/slik-er-

de-norske-klimaskeptikerne

In this paper, «successful PSC» refers to award winning

instances of Popular Science Communication.

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

262

• What role do these DVs have with regards to the

overall message of the texts in which they are

embedded?

o Do they carry core information? Do they

document, illustrate, tell stories? Are

elements of uncertainty represented verbally

or visually?

• What characterizes the interplay between DV and

co-text regarding the construction of identities,

trust and discursive roles?

o Are the DVs related to information

concerning sources and methods? Are there

any invitations for reader activity and/or

dialogue?

• How do the findings related to the questions above

apply to the current norms and expectations of

popular science communication? Do they

contribute to simplicity, engagement and trust?

3 METHODOLOGY

The method suggested to answer the questions above,

is that of social semiotic discourse analysis (SSDA).

This form of analysis belongs to the scientific

paradigm of hermeneutics, linking descriptive,

interpretive, and critical perspectives to generate a

framework aimed at understanding how a certain

semiotic artefact ‘works’ in a certain socio-cultural

context (cf. van Leeuwen, 2005; Ledin & Machin,

2018). In SSDA, the analyst will always relate a close

reading of selected elements on the micro level (the

semiotic structures) to relevant elements on the macro

level (the socio-cultural context) in order to

understand the social functions of the semiotic

choices made by the text producer. Sometimes the

socio-cultural context is investigated directly,

through observation, interviews etc.; sometimes it is

investigated indirectly, through textual implication,

other literary sources or general knowledge. The

approach to the collected data follows an abductive

method, combining the deductive use of certain

theoretical perspectives as a starting point for the data

analysis, with an inductive openness to the unique

features of the investigated objects and their social

contexts (Richardson & Kramer, 2006).

The first stage in a SSDA is most often to gather

a selection of samples from the field under scrutiny

(Aiello, forthcoming). The samples can be described

according to their formal characteristics on a relevant

level of detail. In the next stage, a small or large

number of samples are analyzed with focus on the

realization of the three metafunctions. The number of

samples is dependent on the research questions as

well as the resources allocated to the study. When

dealing with a larger corpus, a pilot-study with close

reading of 2-3 samples can reveal what features that

are most relevant to include in a more formalized and

systematic study of the whole corpus. (cf.

Engebretsen 2017)

The final stage of the analysis includes a

discussion of the findings in relation to the socio-

cultural context that they belong to. What semiotic

and social work do the artefacts do? Do they answer

to the norms and quality criteria of the genre? Do they

represent change and innovation? Do they affect

issues related to power, democracy and equality? This

is the stage where the most important questions are

dealt with, although the answers – as always in

qualitative research – necessarily will involve

interpretations and perspectivations on the side of the

researcher.

3.1 Material for Analysis

In the suggested study, the analyst will follow the

three stages described above, with a focus on the

issues formulated in the four research questions. The

corpus material belongs to the category of “best

practices”. It consists of 20 DVs, found in 20 price-

winning instances of popular science communication.

The price-winners are collected from three different

awards, all relevant to the PUS-discourse.

Both static and interactive DVs will be included

in the sample. The difference between a static

(explanatory) and an interactive (exploratory) DV

may affect all three of the semiotic metafunctions

mentioned in the theory section. E.g.; a static DV may

have a stronger narrative power, while an interactive

DV may lead to a more active and emotionally

engaged reading process.

The selection of samples is obviously not

representative of all DVs applied in PSC. Thus, the

findings cannot be taken as indicative of the total

population. On the contrary, they most probably stand

out from their population, which is the reason for their

status as price-winners. However, this status also

gives them the function of being models for other DV

designers. Close reading of a ‘golden sample’ is

particularly relevant in a genre with such a rapid

development concerning visual and technological

design. What wins prices today, will most probably

affect the mainstream of tomorrow.

What ‘Work’ Can Dataviz Do in Popular Science Communication?

263

ACKNOWLEDGEMENTS

The research reported in this article was supported by

The Norwegian Research Council (NFR).

REFERENCES

Aiello, G., Forthcoming. Inventorizing, situating,

transforming: Social semiotics and data visualization.

In M. Engebretsen & H. Kennedy (Eds.), Data

visualization in society, Amsterdam University Press.

Amsterdam.

Allen, W., 2018. Visual brokerage: Communicating data

and research through visualization. In Public

Understanding of Science, 27(8).

Bauer, M.W., 2009. The Evolution of Public Understanding

of Science – Discourse and Comparative Evidence. In

Science, Technology & Society 14:2, pp. 221–240. DOI:

10.1177/097172180901400202

Cairo, A., 2013. The functional art. An introduction to

information graphics and visualization, New Riders.

Berkeley, CA.

Cairo, A., 2016. The truthful art. Data, charts, and maps

for communication, New Riders. Berkeley, CA.

Davies, S. & Horst, M., 2016. Science Communication.

Culture, Identity and Citizenship, Palgrave Macmillan.

London.

D’Ignazio, C. & Klein, L, 2016. Feminist Data

Visualization. In Proceedings from the Workshop on

Visualization for the Digital Humanities at IEEE VIS

Conference, Baltimore, Maryland.

Engebretsen, M., 2017. Levende diagrammer og zoombare

kart. Datavisualisering som nyskapende fortellerform i

journalistikken. [Living diagrams and zoomable maps.

Data visualization as an innovative form of storytelling

in journalism]. In Norsk Medietidsskrift 02/2017.

Engebretsen, M.; Kennedy, H. & Weber, W., 2018. Data

Visualization in Scandinavian Newsrooms. Emerging

Trends in Journalistic Visualization Practices. In

Nordicom Review 39(2). DOI:10.2478/nor-2018-0007.

Few, S., 2012. Show me the numbers. Designing tables and

graphs to enlighten, Analytics. Oakland, CA. 2

ed.

Fairclough, N., 2012. Critical discourse analysis:

the critical study of language, Routledge. London.

Fleck L., 1979 [1935]. Entstehung einer wissenschaftlichen

Tatsache, Suhrkamp. Frankfurt

Halliday, M. A. K., 1978. Language as a social semiotic.

The social interpretation of language and meaning,

Arnold. London.

Kennedy, H., Hill, R., Allen, W. og Kirk, A, 2016a.

Engaging with (big) data visualisations. Factors that

effect engagement and resulting new definitions of

effectiveness.In First Monday, vol. 21 (11)

Kennedy, H., Hill, R., Aiello, G. & Allen, W., 2016b. The

work that visualisation conventions do. In Information,

Communication and Society, vol. 19 (6).

Kress, G., & van Leeuwen, T., 2006. Reading images: The

grammar of visual design, Routledge. London, 2

ed.

Ledin, P. & Machin, D., 2018. Doing Visual Analysis.

From Theory to Practice, SAGE. LA, London.

Pinney, L., Forthcoming. Is Literacy What We Need in an

Unequal Data Society? In M. Engebretsen & H.

Kennedy (Eds.), Data visualization in society,

Amsterdam University Press. Amsterdam.

Richardson, R. & Kramer, E.H., 2006. Abduction as the

type of inference that characterizes the development of

a grounded theory. In Qualitative Research 2006; 6/4,

pp. 497–513.

Riise, J., 2008. Bringing Science to the Public. In D. Cheng

et al. (Eds.) Communicating Science in Social Contexts.

Springer Science+Business Media B.V. Pp 301-309.

Santerre, L., 2008. From Democratization of Knowledge to

Bridge Building between Science, Technology and

Society. In D. Cheng et al. (Eds.) Communicating

Science in Social Contexts. Springer Science+Business

Media B.V. Pp. 289-300.

Tufte, E. R., 2001. The Visual Display of Quantitative

Information, Graphics Press. Cheshire, CT. 2

Ed.

van Leeuwen, T., 2005. Introducing Social Semiotics,

Routledge. Abingdon, New York.

IVAPP 2020 - 11th International Conference on Information Visualization Theory and Applications

264