Our recent word cloud of the Australian Science Communicators’ Conference Tweets has raised a question: what is the role of non-traditional visualisations in modern science communication?
We all know the value of graphs and maps, but what can we achieve with other visualisations, such as word clouds, network charts, flow charts and so on? What about online only motion charts, as seen in Gapminder?
What do you like about these forms? What don’t you like? Does a word cloud or an online only motion chart have a role in an academic journal article? What are your favourite new forms of visualisation? Where do you see the future in visualisation?
For me Word clouds are a fairly useless way of conveying the meaning or intent of a text. As stated on the Wordle site they are a toy with perhaps a certain amusement value but appear to be of little value to scholarship.
The layout uses only one metric (the size of the font) to convey information and ignores many of the other elements available in good graphic design such as the layout of the symbols on the page (which can be employed to indicate, say relationships), and the meaningful use of colour.
Meaningful graphics have two aspects: (a) The representation of information or ideas by an author and (b) the interpretation of a graphic to extract meaning by an audience. For word clouds, given a text one can create a cloud which conveys a limited representation of what the text is about. However the inverse process where one is preseented with a word cloud from which one may wish to extract the meaning of the text is a pretty futile task.
One may even question the value of a simple count of the frequency of occurrence of a word as a useful metric given that the higher the probability of its occurrence, the lower its information content. It is the less common words which give meaning to a text.
In contrast to word clouds, graphics dispalys of social networks (e.g. Netdraw available with UCINET) present a rich representation of a social network with the visual charcteristcis of the nodes representing the attributes of the actors and the ties connecting the nodes showing the relationships between the actors and the attributes of those relationships.
I don’t think anyone would suggest that word clouds offer perfect representations of the meaning of a text – but they do offer another revealing layer through which you can gain a deeper understanding of a text or discussion.
At a glance, a word cloud can show you focal points and absences in a text, and framing concepts that dominate against others.
Meanwhile, a word cloud can partially capture a disparate discussion that can’t by shown via normal, linear text. This is particularly the case in things like the ASC conference word cloud shown above, which is drawn from the tweets throughout the conference. These tweets – almost by definition – would not be read in their entirety by anyone, so instead a word cloud can offer a revealing snapshot of what was discussed in general.
This is also being discussed on sandpaw, the undergrad sci com students’ blog, here http://blog.cpas.anu.edu.au/sandpaw/2010/03/03/communication-without-words/. Worth following up in sci com more broadly?
Word clouds appear to take a highly reductionist, rather than holistic, approach to the representation of a text. The focus is entirely on the individual elements (the words) and their frequency rather than on their relationships which is where meaning lies.
In many ways word clouds are like trying to describe a jumbo jet by listing an inventory of the number of each type of component: the number of nuts, bolts, pieces of wire, sheets of metal and so on. Prestented with such a list there is no way that one could infer that the inventory describes a jumbo jet. Similarly, an inventory of word frequencies (however prettily presented) does little for meaning. In science, we are at last recognising the limitations of reductionist thinking; it would be a pity if science communication were to move in the opposite direction. Rather than simplistic word counts what is needed is an effective way for extracting the meaning of a text and the striking visual representation of that meaning.
@John Rayner: I think that the reductionism of the type you describe is certainly a problem when it replaces other forms of meaning – when, as your example shows, the number of bolts is held as meaningful above the totality of the jumbo jet.
Yet in context (that is, when we know the holistic meaning of the jumbo jet or particular text), abstractive representation can add to our understanding.
To continue the metaphor of the jumbo jet, wouldn’t a chart of all of its pieces show us something utterly fascinating about the complexity of both the total object and its production process? Wouldn’t such things in fact be impossible to show through the holistic, non-abstract jumbo jet?