Michael Friendly recently sent me this Common Sense Revolution visualization by Scott Sørli plotting a timeseries from 1985-2007 of welfare income for a single person in Ontario; and the names of all the homeless who died on the streets of Toronto over the same time period. An inverse correlation is strongly apparent implying a potential causal relation between the welfare amount and the homeless deaths. While the deaths could have been a simple line chart or bar chart, stacked names much more strongly indicate that we’re dealing with people. And more so that a stack of people icons, these are named people: real people with real given names, real surnames and presumably families and connections in their communities, such as Floyd Anderson, Cheryl Lynn Gunn or Norma/n Lewis. And, disappointingly, there are quite a few John Does and Jane Does, where presumably the investigators did not have enough resources to track down the real names of the deceased homeless person.
It’s also a reminder that text visualizations have a long history. In my book, I do look at a lot of historical text visualizations – as a basis for creating a framework for considering the many ways data can be encoded into text. And then given the framework, I create many visualizations.
But it’s also highly useful and relevant to continue to look at historic examples, to see techniques, combinations, and methods that may inspire or inform future visualizations and creative works. I recently found a copy of Language & Structure in North America (November 4-30, 1975, Richard Kostelanetz curator). Here’s a few interesting snaps of visualization-like uses of text from the 1970’s:
Leftmost is portion of George Maciunas‘ The history of Fluxus, a text-centric flow chart organized by time indicating in historical art movements leading up to Fluxus. The polar plot is an analytical diagram by Agnes Denes titled Studies of Time/ Exploration of Time Aspects, plotting concepts vs time past/present/future further organized by dimensions such as memory – a prioi knowledge, and reproductive – modification. Noise Text #1 by Ascher/Straus is a result of a series of transformations on texts into what appears to be a set of textual vectors.
Visualizing prosody isn’t new. Here’s a great example from 1969 by Ernest/Marion Robson, using letter width to indicate duration, and font-weight to indicate intensity as well as a baseline shift. Not surprisingly, the encoding is very similar to the example visualizations which I’d created as these are connotative mappings. I like their much more dramatic variation in width and use of all caps, overplotting, and use of leaders (….) and whitespace (from Introduction to Transwhichics, DuFour Editions, PA, 1969):
And here’s a very interesting creation of a 3D visualization based on an analysis of syllables per unit measure from Yeats by Beth Learn 1975 (Timeslide Over/Time):
The final two examples are generative works, creating new text from pre-existing work. On the left, a receipt is used as the basis for constraining words by Karen Shaw titled $8.40 (1975) (did not find a good link for Karen). Each line item on the receipt sets the cost per word, where each letter has a unique cost. Words are then stacked into two alternative poems:
On the right, John Perreault, Goddess, 1969, uses parentheses to mark words within larger words or spanning across words, e.g. “(Eve)n in(to t)h(in)e own (so)ft-(con)che(d ear):” thereby creating alternative readings.
Creating and understanding alternative texts becomes more important with an increase in computational textual analytics. Whether overlaying analyses such as attention or assessing generative text sequences, these artistic approaches hint at some possibilities for visualizing text.