About 6-7 years ago, I sat it on a memorable conference presentation from an older gentlemen who read out loud (from his long printed sheet of paper, I must add) a list of PG-13 words of sexual body parts in Madame Bovary, their frequency of use, and placement in the novel’s narrative. We all anticipated the scene of the long carriage ride, especially, and giggled as the man read off the words and numbers in a monotone voice. That voice comes back to me when I’m introduced to tools like Voyant or NGram. I’m struck again by the mix of base anticipation and giddy pattern validation. As readers with a better notion of context, we’ve mapped a great deal of meanings across our favorite texts, and perhaps still chuckle at a computer telling us what we suspect.
Yet, I think Brett (2012) brings a good point when she describes topic modeling as a great tool for discovery, but not necessarily evidence. Evidence was my first assumption for its use, though I’ve always been hesitant to put too much weight on generated patterns considering the range of a corpus, context, emphasis on the things unsaid, etc. However, for the moments the patterns do surprise, the tool becomes a great starting point for further investigation.
I’ve thought of ways to use textual-analysis tools as starting points for participant dialogues. Similar to the local mapping project I’m doing for this class, my diss looks into how a cultural group distinctly views place/space—how does that ideology affect place-making, more specifically? I started with the cultural group’s documents, minutes and correspondence surrounding the building of a cultural center from 1981-2001. I wanted to run it through a textual-analysis tool—as only the starting point—because it would ethically come down to the cultural group validating and discussing the meaning behind the data. However, patterns from the corpus of documents became a great artifact and catalyst for discussion with the group. It became a great opening for community members to say, “Huh, that makes sense because…” or “No way! What we really meant was…”
If I were to use it for my LGBTQIA+ map of Lansing which collects audios of locals, I wonder how useful it would be to compile patterns in regards to how they discuss lost sites—and how they might respond to or make meaning from that.