Emotion and Modifier in Henry Rider Haggard’s Novels

Salim SazzedACL 2023

TL;DR

This paper applies computational linguistic techniques to analyze the emotional content and specific grammatical features in six of Henry Rider Haggard's adventure novels. The study reveals a consistent prevalence of "joy" as the dominant emotion and a notable uniformity in the use of intensifiers and mitigators across the novels, offering a novel quantitative perspective on literary analysis.

What Problem Does This Paper Address?

The paper addresses the growing interest in using quantitative methods, particularly Natural Language Processing (NLP), to analyze literary texts. While digital humanities has seen a rise in such approaches, there's a need to apply these methods to understand the nuanced emotional landscapes and linguistic patterns within specific authors' works. This study aims to bridge this gap by investigating the affective content and the deployment of specific linguistic features (modifiers, deontic modals) in the celebrated novels of Henry Rider Haggard, a pioneer of the "lost world" genre. The goal is to uncover patterns that might be missed by traditional qualitative analysis.

Methods

The researchers employed a multi-pronged approach using NLP techniques. For emotion analysis, they utilized the EmoNet framework, which identifies eight primary emotion categories (joy, anticipation, surprise, trust, anger, disgust, fear, and sadness) at the sentence level. For linguistic analysis, they focused on three types of modifiers: intensifiers (words that strengthen meaning, e.g., "very"), mitigators (words that soften meaning, e.g., "somewhat"), and deontic modals (verbs expressing obligation or permission, e.g., "must," "should"). They also analyzed words that frequently co-occur with these modifiers. The text was tokenized into sentences, and very short sentences were excluded.

Datasets

The study analyzed six literary works by Henry Rider Haggard, obtained from the Project Gutenberg library in UTF-8 format:

  • King Solomon's Mines (KSM)
  • Allan Quatermain (AQ)
  • The Holy Flower (HF)
  • The Ivory Child (IC)
  • She (SHE)
  • Ayesha, the Return of She (ARS)

The dataset was preprocessed by manually removing metadata and segmenting the text into sentences using the NLTK tokenizer.

Key Insights and Ideas

  • Joy Reigns Supreme: Across all six novels, "joy" consistently emerged as the most prevalent emotion at the sentence level, accounting for nearly half of the emotional expressions. This aligns with the adventurous and often triumphant nature of Haggard's narratives.
  • Modifiers are Consistent: The study found a remarkable consistency in the use of intensifiers and mitigators across the novels. Intensifiers were generally more prevalent than mitigators, and their usage percentages were quite similar.
  • "Very" is King (of Intensifiers): The intensifier "very" was the most frequently used across all novels, and its co-occurring words (like "good," "well," "much") also showed significant similarity.
  • Mitigators are a Pack: While "very" dominated intensifiers, mitigators like "quite," "rather," and "almost" showed more comparable prevalence, suggesting a more distributed use of softening language.
  • Deontic Modals are Ubiquitous: Deontic modals were consistently present across all novels, with "could" being the most frequent, followed by "should."

Key Results

  • Emotion Distribution: Joy constituted 41.7% to 48.5% of dominant emotions across the novels. Fear, anger, and sadness were the next most dominant, but at significantly lower frequencies (10%-13%).
  • Modifier Prevalence: Intensifiers appeared at rates between 0.182% and 0.322%, while mitigators ranged from 0.139% to 0.278%.
  • Deontic Modal Prevalence: Deontic modals were present in 0.61% to 0.80% of sentences across the novels.
  • Linguistic Consistency: The findings indicate substantial degrees of consistency in the emotional and linguistic attributes studied across all six novels, suggesting a stable authorial style in these works.

Limitations and Future Work

The authors acknowledge that this is a preliminary study. Limitations include the focus on only six novels and the use of broad emotion categories. Future work could involve:

  • Fine-grained Emotion Analysis: Investigating sub-categories of emotions and tracking their changes throughout the narrative.
  • Expanded Linguistic Features: Incorporating a wider array of linguistic features for a more comprehensive analysis.
  • Broader Authorial Scope: Analyzing additional novels by Haggard from different genres to assess consistency and divergence in linguistic and psychological patterns.
  • Relating to Narrative Elements: Exploring how identified emotions and linguistic features correlate with plot twists and other narrative devices.

Summary updated

BibTeX
@inproceedings{sazzed2023emotion,
  author = {Salim Sazzed},
  title = {{Emotion and Modifier in Henry Rider Haggard’s Novels}},
  booktitle = {Annual Meeting of the Association for Computational Linguistics},
  year = {2023},
}
Figure