Type: Other

Text Analysis Toolkit

Mental health and news media test case



The Challenge

There is widespread interest in using data mining techniques to track public opinion by analyzing large volumes of text. Can we leverage some of these techniques to find patterns in how media narratives influence—and respond to—discourse around specific social issues?


The Approach

To address this and related questions, we built a text processing pipeline for broadcast news closed-captioning and initiated a pilot study around the issue of mental health. As a complex and often stigmatized topic, media portrayals of mental illness have attracted the attention of scholars, mental health professionals, and advocates. For our test case, we chose to look at how and when bipolar disorder has been mentioned in broadcast news contexts over a five-year period, paying particular attention to linguistic features such as frames. We chose to use a combination of manual and automated methods to perform our content analysis.


Our Vision

Our research findings will directly inform ongoing refinement of our text analysis toolkit, as well as provide important insights into how a particular topic has been communicated through mass media over time. We plan to use these tools to explore narratives around other issues of interest, and to detect previously elusive signs of cultural change.