Newspapers, Serials, and Journals (Medical)

This topic model is set at 500 iterations, and represents 100 volumes of medical periodicals dealing with shellshock.

To isolate the text for this model into a workset. I used HathiTrust’s advanced search system to isolate newspapers, journals, and serials which dealt with shellshock. I then sorted out all journals which were medical in purpose, and created two collections, one for newspapers which were not medical journals, offering me a collection of 192 volumes, and a second collection of 101, restricted to solely medical journals.

I then exported each collection to a workset, which I accessed in the HathiTrust Analytics Research Center. Once I validated these collections, isolating out volumes which were not open access, I was left with 100 volumes in the medical journal list, and 189 volumes in the general newspaper list.  This topic model only deals with the medical journals.

When set to isolate topics at the k-20 level, the green clusters represent clinical terminology, such as “patient,” “symptom,” “condition,” “infection,” and “diagnosis.”

This orange cluster references military terms, such as “camp,” “lieut,” “fort, and “duty,” which is an interesting word in this context, which would signal to me that I should look further into this corpora, since often, shell-shocked men were seen as shirking their duty.

Expanding out into the k-40 topic range, we see the incorporation of words which may denote race, such as “white” and “colored,” which may be marking the racial background of patients. However, there is also the first introduction of the word “influenza,” which hints at another noise in the data. The First World War, of course, was followed by the Influenza epidemic of 1918, which swept already-devastated Europe.

In this lavender topic cluster, alongside words we might expect to see in a methods section, like “method,” we see “experience,” “group,” “age,” “study,” and intriguingly, “mental.” If these reports are recounting research on mental illness, like shell-shock, this would be worthy of further research.

As we expand outwards to the furthest level of the cluster, to the k-100 level, we see words such as “temporary,” “disease,” “army,” and “hospitals,” alongside the possibly neutral valence of “methods and “tests.”

This topic cluster invites further research into how one might study these articles to highlight questions of health, military history,  disease, alongside questions of race and gender.