From these previous posts, I hoped to show a few things. First, I wanted to showcase how using various text analysis platforms, paired with some closer reading of these texts and some manual searches, allows for much richer investigative experience. Word Clouds and Phrase Nets, despite their drawbacks, when used effectively across various text analysis tools allow us to quickly visualize and formulate research questions. Moreover, they can even help us devise some preliminary findings.
Second, I aimed to try to showcase some of the ways in which the Globe Stories were structurally different from the Public Submissions. Sociologist Arthur W. Frank distinguishes between three different types of narratives people formulate when dealing with a trauma in The Wounded Storyteller (1995). Among them are restitution narratives. Restitution narratives consist of a three part structure from healthy to sick, culminating in a hopeful and happy return to health or “normal” (in his case he was looking at those struggling with severe illnesses). From some of the conclusions mentioned in the “Where are the Bombers?: What Can Word Clouds Tell Us?” and “#BostonStrong”, I think we can classify a significant number of stories as following this general structure. Especially in the “Boston Strong” stories, one can see this narrative pattern, particularly the final section stressing a hopeful return to health–but here, health of not only the individuals but also of the city as a whole. The Globe Stories are a bit harder to classify, and perhaps this is due to their brevity. These stories, on the whole, are shorter than the their Public Submission counterparts. Many of them appear to more like abstracts than full stories, which could account for some of the differences between the two sub-corpora.
These posts have shown me a lot about what kinds of questions and conclusions you can derive from analyzing text. However, I do have some criticisms of my own work. First, which I might have discovered a little too late, I do not think my corpus was substantial enough to draw definitive conclusions about these stories. The Our Marathon archive has been collecting stories for a little less than eight months at this point, so I decided to take a look at everything we had. I had spent a lot of time learning how to try some topic modeling with MALLET, thinking that it would be a substantial addition to my posts. Instead, I realized that no matter how I tweaked the model, the topics that I generated did not reveal any really great insights. At that point, I decided to focus on my other sections.
That being said, I still think this series of posts has its value, not only for the preliminary findings I was able to extract, but also as a building block for future analysis with the Our Marathon material. Our Marathon is continuing to grow and we are planning quite a few outreach days to gather more submissions to the archive. As we gather more material, I can re-apply these methods to check and see if my hypotheses remained. Moreover, as we incorporate more stories, I hope to re-visit some of the questions that my preliminary investigation revealed.
Also, the Our Marathon team is currently involved in a very substantial oral history project with people directly involved in and influenced by the marathon bombings (i.e. survivors, first responders, and local business-owners). And as these audio files are uploaded into the archive, we are making sure they are accompanied by full-text transcriptions. Once these transcriptions are in the archive, I could incorporate that text into my analysis and compare oral versus written narratives of experiences during the marathon bombings. Who knows? I might finally be able generate a meaningful and insightful topic model about the Our Marathon stories.
TABLE OF CONTENTS