Narratives and Trends

Introduction

For our project, we decided to focus on trends surrounding campus location mentions in the Carletonian (Carleton College’s school newspaper). We focused on sentiment analysis, by attempting to understand how emotional associations related to these buildings have changed over time. We chose this project idea because of our group’s interest in how student journalists have portrayed the Carleton college campus. Our data consisted of over 3000 Carletonian issues from 1877-2025. Through trial and error, we created a data visualization of our found trends, which you can find on this page. We recommend opening our graph in another tab so you can follow along with our findings—and maybe even find some interesting trends of your own on the way!

Our Hypotheses

Before we looked too closely at our data visualization, we made some predictions about what we thought we would see based on what we knew about Carleton’s dorm buildings. For one, we thought that the sentiment scores of buildings that had been torn down, such as Gridley Hall and American House, would fall around the time they were demolished, then mentions would cease altogether. We thought newer housing arrangements, such as Cassat Hall, Memorial Hall, interest housing, would be scored higher, while older buildings would be scored lower in recent years.

New Versus Old

Torn Down Buildings

Correlations

We also noticed that some of the lines were very tightly correlated. For instance, the image below shows the lines associated with Myers, Musser, Goodhue, Watson, and Cassat; they continuously stay close together, even during peaks and dips.

Upon looking closer, we realized that this is because many of these dorms are mentioned within the same contexts. For instance, below is an image of one of the pop ups associated with Cassat—you can see that Cassat in mentioned alongside Goodhue, Watson, and Myers, which would account for their tight correlation. If these buildings are being written about in the same sentence, our code would have analyzed similar or even the same contexts to determine a mean sentiment score for that issue.

This led us to our next question: why then, are these dorms being mentioned together?