Fall 2026

Emotional arcs and a new reading of your novel

Due Friday, November 6, 2026, at midnight (11:59 p.m. Eastern, end of Friday).

Use computational analysis and close reading to develop at least one insight about your novel that you had not considered before. Your graphs and selected passages should support an argument about the novel.

Sentiment analysis notebook on GitHub → Open in Colab, or continue in your saved copy.

1. Compare sentiment models and smoothing

Compare the three sentiment methods: sentimentr, TextBlob’s Naive Bayes classifier, and CardiffNLP’s transformer. Explain briefly how they determine sentiment. Overlay all three arcs at 5% and 10% smoothing and include both graphs. Identify a meaningful agreement and a divergence, grounded in passages. Explain what changes with the smoothing window. You may also explore Savitzky–Golay smoothing.

Choose the model and smoothing setting that best support your reading. Justify your choice with textual evidence, including a region where the models disagree. Agreement with your prediction, or with another model, is not enough.

2. Analyze six crux regions from your chosen model

Using your chosen model and smoothing setting, select three peaks and three valleys. For each:

A smoothed point represents an averaged region, not one sentence’s emotion. Read around its approximate center—at least the five sentences on each side provided by the notebook, and farther when needed. Include some regions where the other models disagree. The notebook’s earlier three-agreement/three-divergence exercise is preparation; for this submission, organize your six selections as peaks and valleys from the model you chose.

3. Compare with your original drawing

Include the hand-drawn arc you submitted in class. Compare at least two specific moments with your computational arc. What did the model miss? What did you miss, overlook, or remember differently? Discuss both the overall shape and particular cruxes.

4. Develop a new insight

Make an interpretive claim supported by the overall emotional arc and close reading of one or two crux regions. Explain what you understand about the novel now that you had not considered before this analysis. Describing a tool’s limitations alone does not fulfill this requirement.

Write-up and submission

Write your analysis in a Google Doc, organized into the four sections above. Embed the graphs, drawing, and passages beside your discussion so the reader can follow your evidence.

Include:

Submit two links: your Google Doc and your completed Colab notebook, with outputs saved. Give Professor Elkins access to both links.

Use evidence from your own runs and explain it in your own words. Be prepared to explain a selected graph or passage in class.