Voyant & Week 1

The posts from last week are all up; it’s been a getting-to-know you kind of week. Let’s take a step backwards and see what you’ve all been saying..


Now the word cloud is a much-maligned piece of basic text analysis. Remembering that it’s just a histogram with a bit of style, it can be useful, as Ian reminds us here. Adam Crymble carried out an experiment to see what historians could read from word clouds of text, with interesting results & thoughts on close/distant reading. He suggests, ‘it would be far more useful to have an idea or concept cloud than one that focuses on individual tokens.’ Tokens = words (more or less).

Let’s look at some other aspects of what you wrote. ‘digital’ appears 104 in the 9 posts, ‘history’ appears 114 times. We could imagine that, given the nature of the course, those two terms are going to appear in lock-step. Sara’s ‘wanted’ and Tyler’s ‘it starts’ posts show the term coinciding perfectly, but not so much in others, though the overall trend is roughly parallel (in places). Here, the visualization is not so much about justifying any sort of interpretation, but rather forcing us to look at the posts again in a new light.


Finally, let’s take a look at the networks of named entities in the texts. Voyant can extract people, places, and organizations, and stitch it into a network. One could then analyze this network if we wanted with something like UCINET or Gephi. More on this anon! But what patterns do you see?


Explore the corpus for yourself here: http://voyant-tools.org/?corpus=1389661449112.4785 Remember to filter out stopwords (cogwheel icon).