NLP Analysis of Power and Danger in YA Literature

Description: This research explores the use of power and danger narratives in African American Young Adult (YA) literature, comparing novels by Black authors with those by white authors to uncover biases in narrative styles. Using the power-danger framework by Danforth and Dodds, we analyze word usage patterns to reveal shifts in power dynamics and societal attitudes. By leveraging Natural Language Processing (NLP) tools, this study highlights the role of language in reflecting and shaping perceptions, contributing to both literary analysis and advancements in NLP.

1. Initial Data

Dataset
- 29 award-winning African American Young Adult (YA) novels published between 2010 and 2022
- 27 novels written by white authors, published between 2007 and 2022

2. Research Questions

How does word usage change over the course of multicultural YA novels?
What patterns of bias exist toward low-danger words?
Can meaningful changes and patterns in word usage be found within African American YA novels?
Do power and danger occur differently in YA African American novels than in YA novels by white authors?

3. Pipeline Development

Cleand and tokenized folders of raw txt files.
Tokenized and scored novels using Power-Danger Framework.
Created summary statistics to look for correlation between African-American and White novels.

4. Data Visualizations

Generated time series to show power-danger score fluctuation over time.
- denoised using empirical mode decomposition & internal mode functions.
Generated ousiograms to summarize the distribution of words in a novel.

5. Key Findings

African American authors used more powerful-safe language, emphasizing themes of resilience and empowerment.
White authors used more weak-dangerous language, highlighting themes of external conflict or personal vulnerability.

6. Framework and Methology links

For privacy reasons, I am unable to share code or provide an in-depth description here.