The terms examined in the following discussion of text analysis visualizations include: slaves, master, man, children, people. These particular terms were chosen for analysis in order to explore ways in which the potential humanization of enslaved peoples may (or may not) be found within the corpus of slave narratives.
Unsurprisingly, utilizing the links visualization below, one can see significant connections between the terms “master,” “slaves,” and “man.” This is to be anticipated given what we know about most slave narratives and the relationship between masters and their enslaved persons. What is somewhat surprising is the apparently complete lack of linkage between our other two terms, “children” and “people” with the first three. A closer reading of the incorporated corpus would be needed to more clearly understand this visual disconnect within the graphic. It could have been anticipated that some linkage between the master and children would be apparent through the visualization whether it was in relation to the dynamic of the master with his own children and interactions with enslaved women who served as caregivers, or in connection with the master’s possession of enslaved persons from childhood. This seeming lack of connection deserves further exploration.
The above Mandala visualization shows the relative “magnetic” connections between our selected terms and the corpus of slave narratives. This visualization would appear to confirm our expectations regarding the relative pull of the terms “master” and “man” within the corpus resulting in a weighting of the corpus materials toward those term. At the same time, the mandala appears to show a relative weakness in the connection between the term “slaves” and the corpus materials, especially when placed in direct comparison to “master.” What is less surprising, especially after our interpretation of the links visualization, is the relative distance the term “children” maintains from the entirety of the corpus. Given the lack of connectivity within the previous visualization analysis, this perhaps would be expected.
Knots…no, Bubble Lines…and Knots again
It was my original intent to attempt to utilize the Knots visualizations as my second selection, however, when I entered the selected search terms and ran the text analysis, the initial visuals were both visually not interesting and seemed to show little for further exploration.
As such, I scrapped my use of knots in favor of bubble lines.
As we can see, our selected terms appear rather sparsely in many of the documents within our corpus, particularly those dated earlier. As we scroll down through our documents, however, one clearly notices the increase in prevalence and density of term bubbles beginning with 1789-equiano2, but especially with 1837-ball and 1861-jacobs.
Analysis of the bubble lines visualization will enable us to focus our close reading efforts on those documents within the corpus in which our selected terms appear more frequently and are more closely distributed in relation to one another. Those documents in which our terms appear less often would be given a lower priority when it comes time for closer examination.
Which brings us back to knots…during analysis of the bubble lines, the scarcity of the selected terms as they appeared in several of the earlier documents seemed to correspond to what I had been seeing in the knots visualizations but was failing to recognize. This prompted me to rerun the knots analysis focusing on some of the documents in which the selected terms appeared more prevalent and densely within the bubble lines. The resulting visualizations in knots were far more visually interesting and engaging in relation to textual analysis.
As with the bubble lines visualization discussed above, analysis of the knots visualizations can better focus any follow on close reading of the corpus documents to maximize the effectiveness of our research time in the archive and allowing us to more thoroughly analyze those documents in which the relationship between our selected terms and the potential answers we are seeking may more likely be located.