For this digital text analysis lab, I started by looking at both Sam and Hailey’s side texts in Lexos. I found that, for me, the BubbleViz in Lexos was the easiest way to initially notice word patterns because the text is all right side up, easily legible, and the size of the bubbles make it easy to see what words are being used the most.
When I put these side texts into the BubbleViz, (after taking out stop words), this is the image that I got.
You can’t see very clearly in the image (no matter what I did the picture would not save clearly), but the largest bubble contains the word “goes”. I thought this was interesting, so I went over to Voyant to see what else I could find out about “goes.”
The first thing that I did in Voyant was put the entire novel’s side text in for analysis. I left the stop words in this time and found out that in addition to “goes,” “go” was also used quite frequently. When I looked at “go” and “goes” and their frequency within the novel, I got this graph.
This one actually surprised me quite a bit. Again, this is with the whole novel’s side text, so 1 on the graph is the start of Hailey’s story, 5 is about where her story ends and you flip to Sam’s, and then 10 is the end of Sam’s story. It didn’t surprise me that at 10 there was an increase of “go” and “goes” (which signify death, usually a number of people that die is some incident) because at the end in both of their narratives, there are a lot of things dying and freezing as they climb this mountain. What did surprise me, however, was that at 5, where Hailey’s story ends, there was no visible increase. Looking back on it, this is probably because Hailey’s side text ends as we get into dates that have not actually happened yet, though it was still surprising, nonetheless.
After looking at this, I was interested to see how the frequency of “go” and “goes” shifted in the individual side texts of Sam and Hailey. I expected to see something similar, but this is what I found.
Fig. 3 Fig. 4
You might expect that Fig. 3 would be Sam’s and Fig. 4 Hailey’s because, Fig.3’s frequency steadily goes up while the other is more constant. It is actually the exact opposite. Sam’s frequency is shown in Fig.4, while Hailey’s is Fig. 3. I found this incredibly shocking, especially considering the previous graph that showed a higher frequency at the end of Sam’s novel. So, Fig. 3 shows that at the beginning of the Hailey’s novel, there is a low frequency of “go” and “goes,” (low counts of death), but by the end of her narrative the count is much higher. Unexpectedly, Sam’s novel starts out with a middle-ground number of “goes” and “go” deaths, and though this number fluctuates, it doesn’t show an upward trend really at all.
So, what does this all mean, I have no idea. It might be that Sam’s ending has less death because there is about to be a rebirth with Hailey, there are a lot of different things it could mean. I’ll have to look further into the texts, probably do some more close reading on the narrative portions of the ends of Hailey and Sam’s stories and see how, and if, this “go” “goes” death trend correlates, but it’s definitely something interesting to think about moving forward.
Using these web tools for digital literary analysis was an interesting experience, to say the least. I actually had quite a few problems with using the programs, though it didn’t really have anything to do with faulty software. It was more that it just took so long to try to find correlations and patterns in the text, even though these programs are designed (and ultimately do) speed up the process. Since I used Voyant mainly for these inter-text comparisons, my grievances are mostly centered on its use. The website itself I think is great. It allows you to see patterns in the text that there’s no way (feasibly) you would be able to see my simply reading a text the traditional way. My main problem was that, although Voyant itself is pretty easy to use once you get the hang of it, unless you already have an idea of what you’re looking for, attempting to find a pattern can take a tremendous amount of time. I loved that you could compare word usage and frequency across the text with the graphs, but trying to find a pattern that actually had relevance was more difficult, and I found the checkboxes that you used to compare words that were in the “words in the entire corpus section” on the bottom left of the site annoying. It was difficult to find words using this little box, and if you forgot to uncheck the box before moving on to comparing other terms, you had to scroll through everything to find it once again. Something that I found helpful was to favorite some of the words I thought would be of interest, limiting it down to twenty or so which made them a little bit easier to manage. Overall I think the system is great, but only if you already have an idea of some patterns you want to explore. If you’re just looking to try to find some patterns as you go, you’ll probably be looking for quite a while. I also think that the complexity of Only Revolutions made the pattern finding process more difficult. Because each word, even the little ones like him, her, go, goes, my, and me are so carefully put into this book, I found that eliminating the stop words left out a lot of possibilities for analysis. Because I had to keep these stop words in play, I think Voyant was a little more difficult to manage. If we were using the program to analyze a simpler text, like Peter Pan or some of the other examples I saw in the Liu examples, I think it would be easier to manage.
I actually think that Katherine Hayles’ chapter “How We Read” where she talks about hyper, close, and mechanical reading, and how the three can work together, reflects my opinion on using Voyant and Lexos pretty well. On page 73 she says, “Close and hyper reading operate synergistically when hyper reading is used to identify passages or to home in on a few texts of interest, whereupon close reading takes over. . . Hyper reading overlaps with machine reading in identifying patterns.” I completely agree with this assessment that we should use close reading and hyper reading (in this case, hyper reading through the use of machine generated Voyant and Lexos tools) together in order to better understand a text. By using this machine version of hyper reading, the Voyant and Lexos tools were able to show some patterns in forms such as the BubbleViz and graphs that I would have never been able to find myself by simply reading the novel. Because these are machines and not humans, though, they can’t do close reading of these patters, meaning they can’t tell me what the patterns mean with regards to the text as a whole. This is where close reading comes in. Once I use Voyant and Lexos to identify these patterns (like the “go” “goes” one that I found), it’s then my job to use human close reading to try to figure out what this means. Machine help can only do so much, and though it’s a great tool as a starting point in identifying patterns in the text, close reading of the text and human analysis of the machine generated patterns has to follow to create real meaning.