The Gendered Language in Teacher Reviews (GL) project was created by Ben Schmidt of Northeastern University. While he doesn’t give a specific reason for the project, it seems to be a case of ‘let’s see what we can find in this untapped data set and go from there’ kind of project. He does say he has plans to do an analysis at some point, but doesn’t go into specifics of what that might look like.
Schmidt collected his data in the spring 2014 over the course of several months by scraping text from the teacher reviews on the website Rate My Professor. Although he does not share the scraped text from the reviews, Schmidt is very open in how he created this project. He has a link out in the open to a GitHub repository that contains the code needed to create and run the program. A big portion of the work comes from his previously co-designed Bookworm Project. Bookworm Project is, among other things, is designed to facilitate “interacting directly with the texts in massive digital libraries…[and]to allow richer access to [large text] collections than a simple search engine.” In his GitHub repo, the technical skills used and needed to create this web page are Python, Make file, HTML, and Jupyter Notebook.
In Posner’s terms this project is presented with a text analysis and data visualization through his interactive graph. By adding the visualization, the user is able to easily see the differences in the usage frequency of a particular word or phrase between the two sexes. The additional feature of the graph rearraigning the 20 different listed subjects from highest frequency of use to lowest in descending order, lends additional clarity to the data visualization.
As touched on before, the project is considered still ongoing. In 2015 Schmidt wrote, “I’ve been holding it for a while because I’ve been planning to write up a longer analysis for somewhere, and just haven’t got around to it. Hopefully I’ll do this soon: one of the reasons I put it up is to see what other people look for.” There has been no update on his website about when he might perform said analysis.
The “scholarly primitives” that the project uses and enables sampling, comparing, and illustrating. Sampling comes from the corpus that Schmidt created to draw from; while it is an extensive portion of the website, it is only a sample of the 14 million or so reviews that were on the website at the time. Comparing is fairly obvious, since the whole point of the program is to compare the frequency of words used for male vs female professors. But it also allows for an unadvertised comparison between the academic disciplines. For example, the search term “sex” appears almost six times more frequently in the Health Science reviews than it does in the Biology reviews.
There were a few weak points to this project. A significant one is that even though the web page said that the user could enter and compare multiple words or phrases at a time, I was never able to do so. Regardless of web browser or computer, I was never able to compare multiple terms simultaneously, meaning I had to either mentally recall the graph from one of the terms, or toggle back and forth in multiple tabs. Another potential weakness is the age the data. The corpus was created from reviews that are now 4-5 years old. Since 2014-15, there has been a Big push in feminism, especially with the #MeToo and #TimesUp movements, and it is possible that these cultural changes may reflect on college reviews. This might be an interesting comparison on its own, if the distribution of certain words has changed in half a decade. A very strong point of the project is how open and collaborative this project makes itself Schmidt is very transparent and explicit on how it was created, and this can lead to other programmers and Digital Humanists can use to create other interactive libraries, which Schmidt says he wants.