After having gone through years’ worth of blog posts in order to eliminate what we no longer saw as necessary and relevant to our project, I wonder what possible patterns and histories of the department we could have found and been able to tell had we worked with the full initial data set. Computational text analysis can help find other patterns, narratives, and trends in the data that we could not, due to our project’s goals and time constraints.
From this week’s readings, we learned about various text analysis methods you can use to help find patterns, similarities, and quantitative data about your data set. We used word count and certain keywords to find irrelevant posts in the corpus to delete in order to reduce the amount of physical reading and eliminating we would have to do by hand. Text analysis allows us to go through this process quickly, as well as find other patterns while doing so. One example would be if we are narrowing our corpus down to announcement posts text analysis could help us see what other words pop up with announcement in order to see what the most popular are to announce.
These kinds of tools would be useful, beyond just for finding what to eliminate, but for pinpointing when various “buzz words” become more popular, such as Authority, Power, or Trump. Processes, like tokenizing, can read through the corpus and identify when these words are being used. Thereby, we can see when different words and ideas were having a resurgence in the department, as well as what events were going on outside the second floor to cause these posts to be read.
Another way to see some of these potential patterns is through concordances. Concordances allow you to see similar text that the computer claims are similar to each other. If you were to work with concordances, the program could pull out all the instances in which blog post text talked about promotion. This would then allow you to see how often promotion within the department occurred, when those details became available to the faculty, and who was in the early phases of their career earning promotions compared to those in the later phase of the career.
Text analysis helps us digest large amounts of written work by topic, word, phrase, text amount, etc. By so doing we allow ourselves the space to investigate bigger questions of what were the trends and patterns in the writing of our data. This then allows us to think about larger patterns and themes in the world surrounding our data set which influences what that data looks like. Rather than spending our time combing through piles of words to find a theme, the theme can be brought to us to be analyzed in greater detail and in new ways because of the time we have saved in the early research phase of the word. In that way, text analysis can help us think about what kinds of things the department considers “public humanities,” “academic blog writing,” or other religious studies-related topics because text analysis can help us get the data we need for the analysis we want.