final post: last dance with convo

Take a look at what we’ve been chugging away at for the past few weeks!

Introduction

This project takes on Convocation data from Carleton College, since its beginnings in 1962, to its brief intermission from 1968 through 1977, then to its reintroduction in 1978 until the present, to understanding patterns over time using text analysis. We use Voyant Tools to analyze the text found in descriptions of Convo. Our interactive graph allows users to search for a word or words together. We entered the project as historians, with an eye toward looking for trends we had sensed on the horizon, whether through our own time at Carleton, or through primary documents we have consulted. While we want to provide an open tool for discovery, we also want to have some initial guidance, through a brief synthesis of patterns we have observed in our own research. In essence, we wanted to showcase these while building a tool for further exploration.

Process

Computer sources: In order to generate machine readable data, we had to jump through a few hoops. At first all we had was a scanned PDF of the convocation programs dating back to 1962. We performed OCR (optical character recognition) on the document, but still needed to organize the individual programs other than as a single block of text so that they would be readable by the text mining tool Voyant. In practice this meant copying and pasting the text from the PDF into individual google documents for each term. We then needed a way to get Google Drive’s fiddly .docx file format into something we could actually use. We downloaded the files in .txt form and attempted to host them on the web via the domains we are using for the class. As it would happen, Voyant doesn’t like getting that many links, and uploading a long list of downloaded physical files worked much better. After zipping and storing those files so we could all access them, we realized that our data was cluttered and rather hard to read. At the urging of Jason Mittell, professor of American studies and film and media culture at Middlebury College, we went back in and consolidated the programs into years rather than terms in order to declutter the graph. This involved (you guessed it) another round of copying and pasting text into different documents. The result is a clean looking, interactive, embeddable visualization that we are all pleased with. 

Human brain sources: In order to make sense of the machine readable data, we decided to compare it to other sources. This process included finding sources regarding the history of convocation, student reactions, and changes in the convocation committee. We started with sources from Nat and Nicole in the Carleton archives such as an address defending mandatory convos in the 1940s, a short historic summary, and programs predating the convos in our dataset. Another main sources of information were Carletonian articles from Carleton’s digital archive. We took students and other audience members reactions and critiques to different convocations and recognized the types of convocations they wanted to see, so that we could compare this information to the trends shown in our data. We focused specifically on eras where we knew that changes in convo were being made. This includes from 1968-1978, a period of time in which the students revolted against the mandatory, boring convos and Carleton stopped having them on campus for ten years. A lot of the trends are noticeably different after this time. As students and the CSA (Carleton Student Association) became more involved in the convocation committee, there was more feedback and open communication about the convo speakers and themes.

Phase I (1962-1968): Mandatory, faculty-chosen convocations

Phase II (1969-1978): Student revolt, no convocations on campus

Phase III (1979-2020): Student involvement in convocations, open dialogue between committee and campus

How to Use Embed

So how might one use this graph? It’s very simple. Our hope is that the user will feel inclined to ‘browse’ the data by entering words that pop into their heads and trying them out. To do this, all that is required is typing the word into the search box at the bottom. If you want to clear the data to try a new word, just click the backspace key or click on the ‘x’ on the side of the word. We’ve decided to make the five most frequent words visible on the graph to inspire you, but if you want to get rid of them, just click on them as they appear above the graph. You might notice that Voyant will prompt you by filling in words you begin to type. Some of these words will have asterisks at the end. This asterisk indicates that Voyant will count all words that contain that word, in addition to the word on its own.

Word Frequency in Convocation Descriptions, 1962-1977, 1978-2020

What We Found (So Far)

• “Journalist” increases dramatically over time, this is especially noticeable in comparison to the drop of “professor.” 
• While “religion” remains relatively stable, “theology” decreased over time. 
• “Politics” has always been relevant.
• “Literature,” “english,” and “history” have decreased over time, with “history” having a slight spike in 2019. 
• “Concert” has a bell-shaped distribution, centered around the late 1980s.
• “His” has a higher frequency than “her,” although the difference decreases in more recent years. 
• “Latino” has no results in the 2010s, despite having a high frequency in earlier years. There are no results for “latina.”
• “Climate” and “environment” both increase at about the same rate starting in 1990.

Sources

• The Carletonian archives
• Carleton College Gould Library Archives
• Conversation with Kerry Raadt, Director of Events, on March 7, 2020.