Reflecting the Male Gaze: Quantifying Female Objectification in 19th and 20th Century Novels

TL;DR

This study operationalizes female objectification in literature via two quantifiable biases: agency bias, which compares how often male vs. female entities act as agents, and appearance bias, which uses WEAT-based associations between gendered terms and appearance-related words in text-induced embeddings. By extracting gendered entities and applying a semantic role labeler, the authors estimate agentivity, while embedding spaces fine-tuned on novels yield appearance-sense biases. An empirical analysis of 79 late 19th to mid-20th century English novels reveals systematic female objectification, predominantly in male-perspective works, with female-perspective texts showing no significant objectification on average. The framework provides interpretable metrics and potential for evaluating generated text and corpora, though it relies on tools and binary gender assumptions, highlighting directions for methodological refinement and theoretical expansion.

Abstract

Inspired by the concept of the male gaze (Mulvey, 1975) in literature and media studies, this paper proposes a framework for analyzing gender bias in terms of female objectification: the extent to which a text portrays female individuals as objects of visual pleasure. Our framework measures female objectification along two axes. First, we compute an agency bias score that indicates whether male entities are more likely to appear in the text as grammatical agents than female entities. Next, by analyzing the word embedding space induced by a text (Caliskan et al., 2017), we compute an appearance bias score that indicates whether female entities are more closely associated with appearance-related words than male entities. Applying our framework to 19th and 20th century novels reveals evidence of female objectification in literature: we find that novels written from a male perspective systematically objectify female characters, while novels written from a female perspective do not exhibit statistically significant objectification of any gender.

Figures (7)

• Figure 1: Calculating agency bias for a short story by counting occurrences of female agents, male agents, and male patients.
• Figure 2: GloVe embeddings penningtonGloveGlobalVectors2014 for words in \ref{['table:weat-words']}, plotted along two principal components. Fine-tuning the embeddings on the novel Lady Audley’s Secret by Mary Elizabeth Braddon causes "male" words, but not "female" words, to move away from the cluster of "appearance" words. The novel's appearance bias is 1.575.
• Figure 3: Distribution of publication dates for novels used in our main experiment. Most of our novels were published between the Victorian Era (1837--1901) and the start of World War II (1939--1945).
• Figure 4: The distribution of agency bias and appearance bias scores for our novels, conditioned on author gender (F vs. M) and narrator gender (1p-F vs. 1p-M).
• Figure 5: Our two bias metrics are weakly correlated across our novels dataset ($\rho = \text{.104}$).