Exploring Trends in the Carleton CLAP

About Our Project

What is this project?

The Carleton Literary Association Paper (CLAP), created in 2002 by students Nico Vreeland & Chuk Kittredge, is a weekly comedic zine discussing relevant topics on the Carleton College campus. It was originally created to spread student voices of any kind. Jokes, complaints, opinions, you name it: if it was submitted, it would be published. Now, more than 20 years later, it has maintained the core of its mission, even as its spirit has changed with the ebbs and flows of student interests.

The goal of this project is to analyze topics that have appeared in the CLAP over its history using digital text analysis tools.

Why the CLAP?

Our group was interested in researching the CLAP because we were all fans of the publication and wanted to see where it started out. As we dove deeper into our project, however, we realized that studying the CLAP could open doors for us to understand what is (and what has been) important to the community at Carleton.

A few CLAP covers over the years.

Our Process

To analyze text trends:

  1. We visited the Carleton College Archives in Gould Library and read through their collection of archived CLAPs.
  2. In general, we selected 1 CLAP per year from 2002-2020 and 1 CLAP per academic term from 2021-2024 for scanning and further analysis.
  3. We scanned the physical CLAPs into PDFs and ran Optical Character Recognition (OCR) on them in Adobe Acrobat.
  4. We extracted the text from the OCR’d PDFs by following this guide (requires Automator, MacOS only).
  5. We obtained the original Google Slides (in .pptx format) from current CLAP staff for all 2024-2025 school year issues of the CLAP at the time (all of Fall, 7 weeks of Winter).
  6. We downloaded the slides in .txt format to extract the text.
  7. We uploaded the .txt files and the extracted PDF text files to Voyant Tools to analyze text trends.
    • As part of our analysis, we added to the default “stopword” list in Voyant, which is used to flag words to exclude from analysis. We included terms that were frequent, but often not useful to our trends, like “gmail.com” and “carleton.”

To create our visualizations:

  1. We downloaded the pertinent data from the Voyant Tools-created visualizations in tab-separated format.
  2. We copied that data into Excel and edited it to make it more organized and human-readable.
  3. We uploaded the resulting CSV into Flourish to create more interactive, cleaner graphs than Voyant provides.

Below is our final presentation that we gave in class, which has a section that illustrates our process step-by-step.

Notes on Challenges & Accuracy

Parts of the OCR outputs of the scanned CLAPs were often problematic and incorrect. Even though we did not have enough time or expertise to edit the outputs, we decided to use them in our project anyway, because there was still a significant portion of text which was correctly recognized. However, this means that there is almost certainly information or patterns which have been missed in our analysis.

The OCR text output from part of our 2003 Fall scan.
The original scan of that same part.

This is less of a problem for the CLAPs which we had original digital copies of, but it should be noted that our .txt extraction did not capture any text inside of images (e.g. screenshots, memes), which makes up a significant portion of typical modern CLAP real estate.

The .txt output from a page of the 2025 Winter Week 5 CLAP.
That same page in the original Google Slides format.

We also did not scan or analyze all CLAPs that we had access to, so our final corpus is just a sample of the zine’s scope. We attempted to make our sample as representative as possible, but it is possible that certain trends or topics simply do not appear in the issues we have analyzed.

Contributors

Josephine L., Assata D., and Eliza F.