This module introduced the concept and practice of text mining through the online program Voyant. Text mining, as the readings explained, is much more than using algorithms to find and count words; rather, it consists of numerous approaches to tracking, visualizing, and interpreting data based on a corpus of text. In the reading for one of the example text mining projects, Peter Leonard accurately compares it to a kind of close reading, a traditional humanities practice updated for digital tools: “a text-mining technique with origins in computer science actually enables more
close reading — by pointing to the needles in the haystack in the form of articles saturated with a particular discourse.”
Using excerpts from the WPA Slave Narratives dataset, Voyant allows for numerous ways of interpreting prominent aspects of the sources using text mining, the most prominent of which being through visualization. Sinclair and Rockwell define visualization as “transformations of text that tend to reduce the amount of information presented, but in service of drawing attention to some significant aspect,” and Voyant’s prime example is the Cirris word cloud, which shows how prevalent a word is based on its comparative size. Hence, this word cloud of the North Carolina interviews demonstrates that words such as ‘yankees,’ ‘house,’ and ‘marster’ were among the most common words used over those such as ‘work’ and ‘home.’
The Trends tool also works as a similar visualization tool by graphing the presence of specific words or phrases across all (or some) of the corpus. More so than the Cirrus tool, Trends allows the user to easily track where a word is more common in specific parts of the text, which may illustrate certain themes based on where and how it is used (or not used). For instance, the word ‘free’ was less common than some other distinctive words such as ‘plantation’ in most cases, except in some such as the Missouri corpus.
Moreover, Voyant provides tools for contexts of specific word usages which are naturally important for accurately tracking patterns and themes. In addition to summary tools, Voyant keeps track of the sections of the sources from which each usage of a word appears in order to better understand them. For another example, the contexts tool tracking the word ‘whipped/whupped’ shows how often it came up in the North Carolina interviews as well as what topic or issue it was tied with.
Overall, text mining with Voyant can teach numerous things about a text that may not be immediately evident. Text mining saves the individual reader or researcher by effectively analyzing a dataset for trends and themes based on the words of the text; however, it is still up to the humanist to accurately ascribe meaning and weight to those patterns as they show up.
Sources:
file:///C:/Users/lhrod/Downloads/Text_Analysis_and_Visualizationmod7.pdf
file:///C:/Users/lhrod/Downloads/Robots_Reading_Voguemod7-1.pdf