Topics by Paragraph: Equiano and his Predecessors

A NOTE ON CHRONOLOGICAL SUBSETS

I noted above that one way I have coped with the challenges of visualizing this massive corpus is to limit the data to chronological subsets - it would be helpful to describe some of the chronological peculiarities of this dataset here.

The corpus includes works published between 1734 and 1980 and, of course, the works are not evenly distributed between those dates. As such, regular date ranges (such as decades) sometimes produce subsets that are still too large to be managable in Tableau.

A flurry of narratives were published between the mid-1840s and the end of the Civil War. The highest number of volumes published in a single year was nine, in 1855, while the four narratives published in 1857 boasted the largest annual paragraph count at 35,456.

The beginning of the dataset provides a gentler entry point to comparing narratives. Before Equiano, only seven narratives (eight volumes – including the double-volume by Sancho) had been published. Multivolume publications, including those by Equiano, are treated as distinct titles in these visualizations, but any visualization that refers to “author” aggregates data from all titles within the relevant range by that author.

CASE STUDY: THE EARLY SLAVE NARRATIVE

Narrowing the focus of the dynamic sparklines can help quickly compare one work to its chronological neighbors, but there is only so much meaning to infer from aggregate descriptions of a single piece. Even in the case of multivolume works, like Equiano’s, the aggregate measures urge us to look more closely. What are we to make of the fact that the second volume of Equiano sees a drop in violence (1.8% average to 0.8%), a rise in religious language (2.4% to 3.7%), and a tenfold leap in abolitionist language (0.5% to 5.1%)?

Viewing the topic proportions at the level of the paragraph gives us a good idea of where the topics fall within the broader narratives, allowing us to make some structural comparisons across several narratives.

In the graphs below I have kept the Y-axes fixed, from 0 to 1, to show absolute proportions of topics per paragraph, allowing for visual comparisons across topics. The narratives are ordered chronologically from left to right, beginning with Bluett's account, published in 1734, and ending with both volumes of Equiano's in 1789. Internal order or paragraphs is preserved, so we can make inferences about the structure and pacing of each piece.

What was the state of the genre of the slave narrative when Equiano published his account? These charts make several trends apparent:

• LENGTH: Though paragraphs can vary from a single sentence to well over a page, we can nonetheless gather that the first several narratives tended to be quite brief, expanding some with Gronniosaw's narrative before reaching the two-volume novel-length treatments by Sancho and Equiano.
• COMPARING TOPICS: In the narratives by Gronniosaw and Equiano, language related to Christianity far outweighs the language of violence, most of the others appear relatively even. The piece by Norris seems unusual in that violence seems to occur more frequently than language of religion.
• TOPIC CLUSTERING: Aggregating the average proportion of a topic by paragraph obviously cannot describe the patterns they appear in within each narrative. The cluster of paragraphs all relating to abolitionism near the end of Equiano's narrative is significant; this is the stretch largely describing his involvement in the resettlement campaign.
• OUTLIERS: The high concentration of the "ship" topic in Equiano is unlike any of the other early narratives. This topic contains language related to life aboard ship, in all of its capacities. Equiano's narrative provides a landmark early description of the middle passage, and nearly the rest of his life was spent aboard ship, as a slave to an officer in the British Navy and other masters, and later as an officer in the resettlement expedition.